drug screening research papers

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here .

• Drug screening
• Pharmacology
• Get an email alert for Drug screening
• Get the RSS feed for Drug screening

Showing 1 - 13 of 475

View by: Cover Page List Articles

Sort by: Recent Popular

In silico exploration of PD-L1 binding compounds: Structure-based virtual screening, molecular docking, and MD simulation

Abdullah Alanzi, Ashaimaa Y. Moussa,  [ ... ], Ijaz Ali

Identification of drug-like molecules targeting the ATPase activity of dynamin-like EHD4

Saif Mohd, Andreas Oder,  [ ... ], Oliver Daumke

TAPS-tool reveals severe under detection of substance use problems in patients with severe mental illness

Birgit Seelen-de Lang, Cor de Jong,  [ ... ], Eric Noorthoorn

Scalable lipid droplet microarray fabrication, validation, and screening

Tracey N. Bell, Aubrey E. Kusi-Appiah,  [ ... ], Steven Lenhert

Is yearly interferon gamma release assay latent tuberculosis infection screening warranted among patients with rheumatological diseases on disease-modifying drugs in non-endemic settings?

Carlo Foppiano Palacios, Vaidehi Chowdhary,  [ ... ], Maricar Malinis

Cocaine/crack and cannabis use among transgender women in Goiás, Central Brazil

Larissa Silva Magalhães, Kamila Cardoso dos Santos,  [ ... ], Sheila Araujo Teles

A targeted CRISPR-Cas9 mediated F0 screen identifies genes involved in establishment of the enteric nervous system

Rodrigo Moreno-Campos, Eileen W. Singleton, Rosa A. Uribe

Insights into the computer-aided drug design and discovery based on anthraquinone scaffold for cancer treatment: A systematic review

Hui Ming Chua, Said Moshawih,  [ ... ], Long Chiau Ming

The use of artificial intelligence in induced pluripotent stem cell-based technology over 10-year period: A systematic scoping review

Quan Duy Vo, Yukihiro Saito,  [ ... ], Shinsuke Yuasa

E-CLEAP: An ensemble learning model for efficient and accurate identification of antimicrobial peptides

Si-Cheng Wang

The association between HIV pretreatment drug resistance and virological outcomes in children and adults in sub-Saharan Africa: A systematic review and meta-analysis

Ebako Ndip Takem, Christopher Coox,  [ ... ], Emily Kainne Dokubo

in silico proposal for the treatment of non-small cell lung cancer">Mitoxantrone and abacavir: An ALK protein-targeted in silico proposal for the treatment of non-small cell lung cancer

Juan Enrique Faya Castillo, Richard Junior Zapata Dongo, Paolo Alberto Wong Chero, Stefany Fiorella Infante Varillas

Exploring NCATS in-house biomedical data for evidence-based drug repurposing

Fang Liu, Andrew Patt,  [ ... ], Qian Zhu

Connect with Us

• PLOS ONE on Twitter
• PLOS on Facebook

• Open access
• Published: 28 May 2024

Consensus holistic virtual screening for drug discovery: a novel machine learning model approach

• Said Moshawih 1 , 3 ,
• Zhen Hui Bu 2 ,
• Hui Poh Goh 1 ,
• Nurolaini Kifli 1 ,
• Lam Hong Lee 2 ,
• Khang Wen Goh 3 &
• Long Chiau Ming 1 , 4  

Journal of Cheminformatics volume  16 , Article number:  62 ( 2024 ) Cite this article

1163 Accesses

3 Altmetric

Metrics details

In drug discovery, virtual screening is crucial for identifying potential hit compounds. This study aims to present a novel pipeline that employs machine learning models that amalgamates various conventional screening methods. A diverse array of protein targets was selected, and their corresponding datasets were subjected to active/decoy distribution analysis prior to scoring using four distinct methods: QSAR, Pharmacophore, docking, and 2D shape similarity, which were ultimately integrated into a single consensus score. The fine-tuned machine learning models were ranked using the novel formula “w_new”, consensus scores were calculated, and an enrichment study was performed for each target. Distinctively, consensus scoring outperformed other methods in specific protein targets such as PPARG and DPP4, achieving AUC values of 0.90 and 0.84, respectively. Remarkably, this approach consistently prioritized compounds with higher experimental PIC 50 values compared to all other screening methodologies. Moreover, the models demonstrated a range of moderate to high performance in terms of R 2 values during external validation. In conclusion, this novel workflow consistently delivered superior results, emphasizing the significance of a holistic approach in drug discovery, where both quantitative metrics and active enrichment play pivotal roles in identifying the best virtual screening methodology.

Scientific contribution

We presented a novel consensus scoring workflow in virtual screening, merging diverse methods for enhanced compound selection. We also introduced ‘w_new’, a groundbreaking metric that intricately refines machine learning model rankings by weighing various model-specific parameters, revolutionizing their efficacy in drug discovery in addition to other domains.

Graphical Abstract

Introduction

In the realm of modern drug discovery, virtual screening stands as a pivotal cornerstone [ 1 ]. This computational strategy serves as the beacon for researchers, directing them through vast chemical libraries to efficiently uncover potential drug candidates [ 2 ]. As elucidated by Baber, Shirley [ 3 ], there exists a burgeoning interest in consensus approaches tailored explicitly for ligand-based virtual screening. Such approaches are not mere theoretical constructs; they are the culmination of intricate combinations of multiple properties, each contributing a unique facet to the screening process. Among the properties integrated into these consensus approaches are structural, 2D pharmacophore, and property-based fingerprints. Additionally, scores derived using BCUT descriptors, an Eigenvalue-based molecular descriptor [ 4 ], and 3D pharmacophore methods further enhance the screening's breadth and depth [ 5 ]. Consensus scoring enhances data set enrichment over single scoring functions by approximating the true value more closely through repeated samplings akin to multiple scoring functions, improving active compound clustering thereby recovering more actives than decoys [ 3 ].

Exploring the methodologies employed in consensus docking programs, Houston and Walkinshaw [ 6 ] introduced consensus docking as a method to enhance the accuracy of pose prediction in virtual screening by combining the results from multiple docking programs. The study tested Autodock [ 7 ], DOCK [ 8 ], and Vina [ 9 ], finding that while individual success rates for accurate pose prediction ranged from 55 to 64%, using a consensus approach increased this accuracy to over 82%. This method reduces false positives by advancing only those compounds to the scoring stage that are similarly docked by multiple programs, thereby improving the efficiency of virtual screening and the likelihood of identifying viable drug candidates. Consensus molecular docking workflows are regarded as critical methodologies within virtual screening approaches, primarily aimed at enhancing the identification of genuine actives during virtual screening campaigns [ 10 , 11 , 12 ]. But the exploration doesn't halt in consensus docking software.

Additional studies delve into the intricate tapestry of virtual screening methodologies, uncovering both sequential [ 13 ] and parallel [ 14 ] approaches. Sequential approaches, as the name suggests, unfold in a stepwise manner, systematically applying various techniques on a progressively decreasing number of compounds. This meticulous workflow encompasses stages such as pharmacophore screening, judicious application of property filters, followed by docking, culminating in manual selection. In stark contrast, parallel approaches deploy a multitude of methods independently but on a consistent number of compounds. Techniques such as pharmacophores, similarity methods, and docking are executed simultaneously, culminating in a robust automated selection process [ 15 , 16 ]. In a bid to augment virtual screening's precision, researchers introduce a novel probabilistic paradigm. This framework, meticulously crafted to combine structure- and ligand-based screening methods to improve the accuracy of virtual screening predictions by fusing them into robust probabilities of activity, providing a quantitative bioactivity likelihood for compounds, thereby enhancing predictions [ 17 ].

Navigating further into the heart of the virtual screening, a comprehensive exploration of traditional consensus scoring unfolds. Four distinct methods emerge in this domain: Mean, Median, Min, and Max consensus scoring. Each method, while unique in its approach, seeks to compute compound scores, harnessing quantile-normalized scores drawn from various docking programs. Yet, it is the introduction of advanced consensus strategies that truly exemplifies the study's innovation [ 18 ]. The mean–variance consensus and gradient boosting consensus stand out in this study, seamlessly merging advanced statistical models, gradient boosting mechanisms, and intricate algorithms to refine and enhance score computation [ 18 ]. With the debut of machine learning techniques, the introduction of the Deep Docking (DD) method marks the culmination of this research odyssey. This innovative method, fortified with the prowess of artificial intelligence, addresses the challenges posed by the exponential growth of chemical libraries, offering a beacon of hope for researchers navigating the intricate maze of virtual screening [ 19 , 20 , 21 ]. In our recent work, we introduced a workflow that combines four structure- and ligand-based scoring systems to improve the hit rate with a challenge of a narrow range of active compounds dataset. The results showed that the consensus scoring method outperformed separate screening methods, achieving the highest ROC value [ 22 ].

In this study, various protein targets, including G protein-coupled receptors (GPCRs), kinases, nuclear proteins, proteases, DNA repairing enzymes, and suppressor proteins, were explored. We introduce a novel consensus scoring method for holistic virtual screening. This method employs a sequence of machine learning models organized in a pipeline, with weights assigned based on individual model performance using a novel equation. We have developed an original formula, termed “W_new,” which integrates five coefficients of determination and error metrics into a single metric to assess model robustness. Using this pipeline, we comprehensively evaluated multiple molecular targets, scoring them based on docking, pharmacophore, shape similarity, and QSAR properties, which were used to train machine learning models. The selection of the optimal model, based on its assigned weight, enabled retrospective scoring of each dataset through a weighted average Z-score across the four screening methodologies. Additionally, we validated the robustness of these models using an external dataset to assess predictive performance and generalizability. Enrichment studies were conducted to evaluate the efficacy of the workflow.

The datasets for this study were obtained from the PubChem database [ 23 ] and the Directory of Useful Decoys: Enhanced (DUD-E) repository [ 24 ], which were utilized to amass active compounds and corresponding decoys for the selected proteins. IC 50 activity metrics were curated from PubChem, encompassing a range of forty to sixty-one active compounds per protein. Additionally, a substantial collection of decoys was meticulously compiled, numbering between 2300 and 5000 for each protein. To ensure the robustness and reliability of our study, an assessment for identifying and quantifying bias in datasets was conducted, addressing potential biases in active compound selection and decoy distribution. The active compounds were subsequently segregated into distinct sets for testing and validation, as well as for external validation purposes. The molecular structures were neutralized and compound duplication was removed, salt ions and small fragments were excluded. The IC 50 values were further converted into pIC 50 values using the formula pIC 50  = 6 − log (IC 50 (μM)). Stereoisomers were systematically generated due to the presence of compounds characterized by undefined stereocenters within their SMILES representations.

Assessment of datasets for identifying and quantifying bias

In this study, we employed a rigorous strategy to mitigate bias in analyzing active and decoy datasets for each target, bolstering the credibility of our findings. An essential aspect was the incorporation of an external validation dataset, unseen during model training. This, coupled with satisfactory R2 values, enhances the credibility of AUC and other performance metrics, confirming the robustness of our models. Additionally, our methodology deviates from conventional virtual screening practices, which typically maintain a 1:50 to 1:65 ratio of active to decoys [ 25 , 26 , 27 ]. By adopting a more stringent 1:125 ratio, we increase the challenge of accurately identifying actives within the decoy dataset. Notably, these performance metrics primarily facilitate comparative assessments between consensus scoring and other screening methods, demonstrating the superior efficacy and precision of consensus scoring.

In this assessment, we’ve employed a three-stage workflow to validate the datasets, following Sieg and Flachsenberg’s criteria for comparative analysis with MUV datasets to identify differences [ 28 ]. This methodology addresses issues highlighted by Sieg et al., particularly biases arising from uneven distributions of physicochemical properties among active and inactive groups, which can skew model outcomes. We also examined “analogue bias,” where numerous active analogues from the same chemotype inflate model accuracy. This approach enhances structural diversity within the datasets, reducing variability in predictive accuracy and yielding more robust and generalizable machine learning models [ 29 ].

We initially assessed seventeen physicochemical properties to ensure balanced representation between active compounds and decoys for each protein target. Fragment fingerprints were then used to prioritize diversity in compound selection and analyze patterns of similarity and diversity among active compounds and decoys. Two-dimensional principal component analysis (2D PCA) was applied to visualize the positioning of active compounds relative to decoys for each target. To refine the calculation of median active neighbors among decoys, adjustments were made to align with the actual decoy pool size and the 1:125 active-to-decoy ratio. This enhanced the evaluation of spatial relationships within chemical space and improved detection of compound distribution patterns and potential dataset biases. To compare with established datasets, we sampled two random datasets from the Maximum Unbiased Validation (MUV) dataset, maintaining the same active-to-decoy ratio used in our study [ 30 , 31 ].

Calculation of fingerprints and descriptors for active compounds and decoys

In this study, RDKit [ 32 ] open-source scripts were utilized to compute a wide range of molecular fingerprints and descriptors for both active and decoy compounds associated with each protein target. These descriptors encompassed Atom-pairs, Avalon, Extended Connectivity Fingerprints-4 (ECFP 4), (ECFP 6), MACCS, Topological Torsions fingerprints, as well as partial charges. Additionally, a set of ~ 211 descriptors provided by RDKit was incorporated as chemical compound features. For a comprehensive understanding of the specific features employed, the pertinent code snippets are available in the GitHub source repository.

Selection of protein targets and crystal structures

We selected a carefully curated set of protein targets, including nuclear receptors, kinases, and enzymes, for investigation. These targets underwent robust validation using both active compounds and decoy ligands. Additionally, we deliberately excluded a subset of external datasets from the training and testing datasets to prevent data leakage and enable evaluation of the computational models’ predictive robustness. Crystal structures of macromolecular targets (AA2AR, AKT1, CDK2, DPP4, PPARG, and EGFR) were obtained from DUD-E, along with their corresponding sets of active and decoy ligands. Active compounds for TDP1 and the p53 suppressor protein were sourced from PubChem and the scientific literature, encompassing anthraquinones and chalcone chemical classes [ 33 , 34 , 35 ].

To prepare the protein and ligand structures for subsequent analyses, Autodock Tools were employed. Protein crystal structures were retrieved from the Protein Data Bank (PDB) [ 36 ], where hydrogen atoms were systematically added, and water molecules were effectively removed. Furthermore, the dimensions and resolution of the grid maps were established utilizing the AutoGrid tool. All compounds were subjected to docking against the reference receptor, confined within an 18 Å cubic enclosure centered around a co-cyrstalized ligand. Protonation states were computed for all proteins within a pH range of 7 ± 2, with the aim of aligning them with the physiological pH conditions. The redocking procedure was applied to all protein targets with their respective co-crystallized ligands.

Pharmacophore scoring

In the analysis of each of the eight datasets, we conducted an assessment aimed at identifying the most diverse molecules, with the objective of quantifying their resemblance to the remaining compounds within the dataset. Utilizing the RDKit and SKlearn packages, an algorithm was employed to systematically traverse the data rows within the DataFrame. The ECFP4 for each compound were calculated, and these fingerprints were then subjected to K-means clustering using the scikit-learn KMeans algorithm. Notably, the selection of a cluster count within the range of three to five was made to ensure that each resultant cluster would distinctly represent a chemically disparate group. Each cluster was subjected to a superimposition process, enabling the detection of common pharmacophore attributes, guided by a set threshold mandating the minimum presence of 3 to 5 of these features. Pharmacophore features were computed for each cluster using phase module in Schrödinger suite [ 37 ]. Each compound was scored by the group of features calculated in its cluster. This module allowed us to generate a pharmacophore model that encapsulates the essential structural elements required for potent ligand binding. To assess the predictive power of our pharmacophore model, we calculated the Root Mean Square Error (RMSE) for each active compound based on their feature matches with the model. This quantitative measure provided a reliable indicator of the model's accuracy in predicting bioactivity.

Docking scoring

The protein structures were retrieved in PDB format and processed using AutoDock Tools [ 7 ]. Active compounds were formatted accordingly and converted to PDBQT format using AutoDock Tools, which contains crucial ligand property information and is compatible with AutoDock. Ligand preparation involved adjustments for stereochemistry, protonation, and the addition of polar hydrogen atoms using AutoDockTools. Gasteiger partial charges were assigned, and details regarding rotatable bond torsions were incorporated into the PDBQT format. Identification of the protein's binding pocket was based on available structural data or by referencing the binding site with the co-crystallized ligand in the original PDB file. A cubic grid box was defined around this identified binding site, tailored to encompass the pocket adequately while allowing ample space for ligand exploration. Grid spacing was determined at an optimal value (0.375 Å) to balance computational efficiency and precision. Molecular docking involved exploring the optimal conformation and orientation of the ligand within the receptor's binding site. AutoDock Vina was utilized to accommodate flexible ligands, prioritizing conformations and binding interactions resembling those of the co-crystallized ligand to calculate docking scores.

2D fingerprint shape similarity scoring

From each of the eight datasets, we computed the most diverse molecules to evaluate their resemblance to the remaining compounds within the set. The code used RDKit and SKlearn to extract SMILES notations from the DataFrame, compute ECFP4 fingerprints, and perform K-means clustering with K-Means algorithm. To ensure each cluster had a representative compound, the number of clusters was limited to three or four. Representative compounds were determined by choosing those with the longest SMILES notation, ensuring greater complexity and diversity as a selection criterion [ 38 ]. Subsequently, shape similarities between each active compound and the reference compounds were computed using the Tanimoto similarity metric. This script serves to compare a specified chemical reference compound against a collection of additional compounds in a CSV file, quantifying their structural similarities via the Tanimoto coefficient. The highest index for each compound from the reference compound list was considered. Code snippets executed to perform this process have been added to the GitHub link.

Development of the weighted metric (W_new) for evaluating machine learning models

A comprehensive ensemble of twelve machine learning models was employed, each offering adaptable parameter tuning through grid search techniques tailored to the specific requirements of each case. These models encompassed Decision Trees, K-Nearest Neighbors (KNN), AdaBoost, Random Forest, Linear Regression, Elastic Net Regression, Gradient Boosting, XGBoosting, and various Support Vector Regression (SVR) models, including linear, sigmoid, Radial Basis Function (RBF), and Nu-SVR kernels. These diverse models were seamlessly integrated into a unified codebase, offering two distinct options for feature selection: Principal Component Analysis (PCA) or Mutual Information (MI) feature selection. To assess the models' robustness and performance across different cases, we introduced a weighted ranking system based on five key evaluation metrics: R-squared (R 2 ) for training and validation sets, Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and Mean Absolute Error (MAE).

In the proposed composite metric formula, several statistical measures are integrated to comprehensively evaluate the performance of a model. The formula begins with the sum of squared R-values, R 2 _train + R 2 _val, which represents the proportion of the variance in the dependent variable that is predictable from the independent variables, so this sum reflects the total explanatory power of the model over both datasets. When both R 2 _train and R 2 _val are high, their sum, is also high. This sum is part of the numerator in the formula, so a higher sum of performance metrics (P) will contribute to a larger value of W_new [ 39 ].

Additionally, the formula includes the sum of error metrics (E), namely MSE, RMSE, and MAE. This sum represents the aggregate magnitude of prediction errors, irrespective of their direction. These terms form the denominator in the main fraction of the formula. Lower values of MSE, RMSE, and MAE result in a smaller denominator. Since dividing by a smaller number results in a larger value, this will increase W_new [ 40 ].

We computed the absolute difference (D) between \({R}_{train}^{2}\) and \({R}_{val}^{2}\) , and then create an adjustment factor to account for the discrepancy:

Then, we added the adjustment factor to penalize discrepancies between training and validation performance:

We combined the performance metric sum (P) with the error metric sum (E) and adjust based on the discrepancy adjustment factor (A):

Finally, we normalized W to ensure it's within a specific range (0–1), by dividing it by \(1+W\) [ 41 , 42 ].

Putting it all together, we get:

The proposed weight formula, “w_new,” assigns higher weights to models with superior performance, characterized by elevated R 2 scores in training and validation, decreased MAE, RMSE, and MSE values, and smaller discrepancies between training and validation R 2 scores, indicating resistance to overfitting. Conversely, models with lower performance receive reduced w_new values, as evidenced by diminished R 2 scores, increased MAE, RMSE, and MSE values, and larger training-validation R 2 score gaps. Notably, w_new is applicable when training and validation R 2 scores range between 0 and 1, with the algorithm excluding results beyond this interval to ensure the identification of adequately performing models.

Furthermore, it's noteworthy that each individual machine learning model integrated into the aforementioned code was fine-tuned using the w_new formula. This fine-tuning process involved specific cross-validation techniques and the selection of an optimal number of PCA components or features through the script. This meticulous approach facilitated the identification of the best-performing machine learning model, characterized by the highest w_new value. All the code snippets, performed in this study, have been documented and made accessible on GitHub.

Establishing predictive workflow through consensus holistic virtual screening

Upon identifying the most robust model for each dataset using four scoring methods—PIC50 (QSAR), pharmacophore, docking, and shape similarity—a detailed evaluation was conducted. This included training each model on a training dataset and evaluating its performance on a separate validation dataset, split in a 70:30 ratio. Further validation was performed on an external dataset to compute R 2 and confirm prediction robustness. For a holistic model assessment, both active and decoy compounds were scored using the same approach, with scores standardized via z-scoring. Each score was then adjusted by the w_new factor from the previous step. A weighted average score was calculated for each compound, leading to their descending order ranking based on these scores. This ranked list underpinned the creation of an enrichment curve, as depicted in Fig.  1 .

Comprehensive Workflow for the Consensus Holistic Virtual Screening. A Selection of protein targets spanning diverse categories, including G protein-coupled receptors (GPCR), kinases, nuclear proteins, proteases, and other targets. B Calculation of fingerprints and descriptors for both active and decoy datasets, along with the computation of four distinct scoring metrics for active datasets per target. C Integration of twelve machine learning models in the pipeline to identify the most optimal dataset within each scoring category. D Utilization of a novel formula to determine optimal parameters based on the highest w_new value. E Evaluation of the entire workflow's performance, including ROC curve analysis and other metrics, to demonstrate its effectiveness

In this study, we analyzed eight protein targets across diverse functional categories, including GPCRs, kinases, nuclear proteins, proteases, DNA repair enzymes, and tumor suppressor proteins. Table 1 details the meticulous examination of active and decoy compounds sourced from the DUD-E database for each target. Notably, active compounds for TDP1 and p53 were exclusively selected from anthraquinone and chalcone chemical classes, sourced from PubChem, BindingDB, and literature. Decoy sets for these targets were generated using the “Generate Decoys Tab” in DUD-E. This departure aimed to evaluate the efficacy of the consensus holistic virtual screening strategy across diverse datasets. Additionally, the methodology was evaluated for its impact on performance metrics within new settings [ 32 ], building on previous evaluations. External datasets were used for predictive capability assessment, and R2 values were calculated for validation.

Comparative analysis of bias in datasets distribution and diversity

Figure  2 A displays the distribution of active compounds among decoys across each target protein, along with their neighboring active and decoy compounds. Except for TDP1 and p53, distribution patterns across other targets closely resembled those in MUV datasets (particularly MUV-737 and MUV-810). Active compounds were positioned in central and peripheral regions, indicating diverse interactions with other actives and decoys. Deviation in TDP1 and p53 datasets is attributed to their unique composition with anthraquinone and chalcone derivatives, suggesting stronger connections among themselves and differentiation from decoys. These datasets were designed to explore dataset incompatibilities, as previously studied [ 22 ], and their influence on performance metrics was assessed in the current study.

Comparative analysis of active compounds and decoys across eight datasets employed in this study and two MUV datasets. A Similarity maps, generated via the 2D Rubber Band Scaling algorithm utilizing fragment fingerprints, depict the spatial arrangement of active compounds in comparison to decoys. These maps are color-coded according to the diversity selection rank, offering a visual representation of the compounds' distribution. B Distribution of physicochemical properties for seventeen distinct properties between actives and decoys. C Principal Component Analysis maps constructed from eight types of descriptors, demonstrating the segregation of active compounds from decoys

The Rubber Band Scaling algorithm used in the similarity maps assigns compounds random positions in a quadratic space to minimize the distance between them. Optimization cycles adjust compound positions based on their similarity relationships defined by the Fragment Fingerprint descriptor. Compounds are moved closer or further apart to reflect their chemical similarity, ensuring similar compounds are close neighbors in the visualization [ 43 ]. The maps are color-coded based on diversity selection ranking, with higher values indicating less diversity (green) and lower values indicating more diversity (red). Similarity in this metric between active compounds and decoys suggests homogeneity in chemical class diversity. However, greater diversity among active compounds can enhance heterogeneity in training and testing sets, minimizing bias in machine learning scoring functions, as described by Li and Yang [ 29 ]. Refer to the Supplementary Material 3 file for a detailed view of the components in Figure 2 A, B, and C.

Data from the similarity maps, presented in Table  2 , reveal average diversity rank differences between active compounds and decoys across various target datasets. The diversity range of these datasets aligns with that of the two MUV datasets in this study, facilitating comparative diversity analysis against a recognized benchmark. Notably, some datasets, like DPP4, show no significant diversity differences between actives and decoys, while most exhibit significant differences. Unlike MUV-810 and DPP4, most datasets feature more diverse actives (lower values) than decoys, potentially enhancing training and testing compound diversity relative to decoys [ 44 ]. The most pronounced differences in diversity ranks between active compounds and decoys were identified within the TDP1 and p53 datasets, translating the graphical clustering of active compounds into a quantifiable disparity in diversity rank. This distinction does not imply higher overall diversity but rather delineates the active compounds' separation from decoys, attributed to their aggregation in confined areas of the maps.

In Fig.  2 B, seventeen physicochemical properties were computed for all datasets and compared with two MUV datasets. The minimal differences between actives and decoys across the protein target datasets, ranging from 7 to 11, mirror the consistency seen in the MUV datasets, where 10 to 11 non-significant property differences were observed in MUV-810 and MUV-737, respectively (see Table  2 ). This indicates fewer disparities between actives and decoys, enhancing dataset reliability and comparability with established benchmarks [ 45 ]. In the final validation phase, PCA was used to visualize both active compounds and decoys, incorporating all utilized fingerprints and descriptors from model training. Classification was performed to differentiate between active and decoy compounds based on predefined titles, enabling focused examination of molecular characteristics distinguishing active compounds from inactive ones. Euclidean distances between each active compound and all decoys within the dimensionally reduced space were computed, with a threshold distance set by the 10th percentile of these distances facilitating identification and enumeration of decoys considered 'neighbors' to each active compound. This neighbor count served as an indicator for assessing the similarity level between actives and the decoy-dominant chemical space. The analytical results were summarized into a statistical metric, the median number of neighbors, subsequently normalized against the decoy count and the active-to-decoy ratio percentage. For a graphical representation of this process, refer to Fig.  2 C.

As demonstrated in Table  2 , AKT1 exhibited the highest median number of active neighbors among decoys, with a value of 563, followed by MUV-737 at 552.6. Conversely, TDP1 and p53 displayed the lowest median numbers of neighbors, at 112.23 and 74.35, respectively, with MUV-810 showing the third lowest at 318.60. The diminished neighbor count observed for these active compounds suggests a higher selectivity or a lower chemical similarity compared to actives surrounded by a greater number of decoy neighbors [ 46 ]. As previously mentioned, the actives within the TDP1 and p53 datasets belong to two distinct chemical classes, leading to a propensity for clustering amongst themselves rather than mingling with decoys. This distribution highlights how the protein datasets in question align with the benchmark established by the MUV dataset.

Analysis of different screening scores across macromolecular targets

It is noteworthy, as illustrated in Fig.  3 , that the distribution of PIC 50 values of both p53 and TDP1 diverges significantly from the broader spectrum of other macromolecular targets, with the latter target dataset exhibiting a considerably wider range of activities. Additionally, we must highlight the relatively balanced distribution observed across various scoring metrics, encompassing pharmacophore analysis, docking simulations, and similarity scoring for both TDP1 and p53. Of particular interest is the exceptional performance observed in the case of similarity scores, which are distributed more evenly across the entire cohort of targets. In contrast, pharmacophore scores, followed by docking scores, reveal less uniform distributions for specific targets. Nevertheless, it becomes apparent that distinct computational methodologies yield varying levels of performance, not intrinsically associated with their respective average “w_new” values. Among these methodologies, the pharmacophore approach emerges as the most robust, displaying the highest average “w_new” value of ~ 0.965. Closely following, the shape similarity method demonstrates commendable performance, with an average “w_new” value of ~ 0.895. Conversely, the results of Docking screening yield a comparatively lower average “w_new” value, ~ 0.681. Lastly, the “PIC 50 ” scoring approach exhibits the least favorable performance, denoted by its lowest average “w_new” value of ~ 0.671. These findings underscore the considerable variability in the predictive capabilities of these screening methodologies within the context of our study.

Distribution of various bioactivity metrics across different protein targets. The four panels represent the distributions of Docking, PIC50, Pharmacophore, and Similarity values for eight protein targets (AA2AR, AKT1, CDK2, DPP4, TDP1, PPARG, EGFR, and p53). Each violin plot depicts the distribution of values for the respective metric, with the width of the plot at different values indicating the density of data points. The inner lines represent quartiles of the distribution

Machine learning models generated and their performance

In this study, several machine learning models exhibit distinct performance metrics. SVR models that include all kinds of SVR and Nu-SVR models with different kernels, on average, yield an R 2 -training score of ~ 0.854 and an R 2 -validation score of 0.749. Its MAE stands at 0.147, with an RMSE of about 0.180. The Adaboost models achieve an average R 2 -training score of 0.967 and an R 2 -validation score of 0.825. Decision Trees, characterized by a more flexible structure, report an R 2 -training value of 0.843 and an R 2 -validation value of 0.709. The Elastic Net and linear Regression models present an R 2 -training score of 0.878 and a validation score of 0.792. Gradient Boosting, a boosting ensemble method widely used in QSAR modeling [ 47 ], showcases impressive scores with an R 2 -training of 0.999 and an R 2 -validation of 0.978. The k-Nearest Neighbors (KNN) models register an R 2 -training score of 0.999 and a validation score of 0.878. Across these models, the w_new parameter displays a range of values, with Gradient Boosting exhibiting the highest average value of 0.974, suggesting its superior performance in the given context as depicted in Fig.  4 .

Comparative Analysis of Machine Learning Model Performances in the consensus holistic workflow: The upper panel presents a series of box plots showcasing the distribution of performance metrics such as R 2 validation and training, W_new, MAE, MSE, and RMSE for various machine learning models. The lower panel illustrates the R 2 values for external validation of four key predictive features—PIC 50 , Pharm, Docking, and Similarity—across multiple target proteins, providing insights into the predictive accuracy and reliability of the models employed

The evaluation of an external validation dataset reveals variable predictability among proteins, with R 2 values ranging from 0.625 for p53 to 0.891 for AA2AR, reflecting differences in inhibitory concentrations. High R 2 values for AA2AR (0.891) and EGFR (0.797) indicate potent inhibitory effects, demonstrating the models' predictive accuracy. Pharmacophore scores, particularly for AA2AR (R 2  = 0.905) and PPARG (R 2  = 0.810), suggest reliable pharmacophore model predictions. Docking scores vary, with CDK2 (R 2  = 0.766) and PPARG (R 2  = 0.739) indicating precise docking efficiency predictions. The analysis of 2D fingerprint shape similarity metrics shows significant variation, with DPP4 and TDP1 exhibiting higher scores, while p53's lower value is attributed to the dataset's small size, as shown in Fig.  4 .

In the pursuit of robust scoring methods for producing a robust consensus holistic virtual screening within a diverse set of molecular targets, various machine learning models and kernels were employed, each yielding specific w_new values indicative of their performance. The Docking scoring method primarily employed SVR ML models with an RBF kernel, resulting in a w_new value of 0.872. In contrast, the QSAR (PIC 50 ) scoring method utilized the same SVR ML model with an RBF kernel, yielding w_new average value of 0.888. The shape similarity scoring method was predominantly associated with the Adaboost ML model, which produced w_new value at 0.969. Similarly, the pharmacophore scoring method was best represented by the Adaboost ML model, achieving the highest w_new value of 0.986 among all scoring methods screened as illustrated in Table  3 .

Factors influencing w_new values

To find out the factors with a higher influence on w_new and the effects of model complexity against performance metrics we employed several techniques. We analyzed the correlation between the five performance metrics previously mentioned with cross-validation times, number of PCA components/features, and model parameters such as model cost and gamma, Nu value in SVR, L1 (Lasso), and L2 (Ridge) regularization in addition to other hyperparameters according in the model employed as clarified in the Supplementary information Table  1 . From Fig.  5 , The correlation coefficients between w_new and the various metrics are as follows: R 2 -training equals 0.1265, R2-validation is 0.4638, MAE =  − 0.9022, RMSE =  − 0.9324, MSE =  − 0.8729. R 2 -training and R 2 -validation have positive mild and moderate correlations with w_new, respectively. However, the correlation with MAE, RMSE, and MSE have strong negative correlations with w_new. As the error metrics increase, w_new tends to decrease. Among the error metrics, RMSE has the strongest negative relationship with w_new, followed by MSE and then MAE.

Pairplot shows the correlations between performance metrics and models parameters, cross-validation, and numbers of PCA and features components

In pursuit of a deeper understanding of the contributions of various metrics to the variable w_new, a multiple linear regression analysis was conducted. This rigorous examination sought to discern the individual influence of each metric on w_new while effectively controlling for the presence of other metrics. The formulated multiple linear regression model is articulated as follows:

The multiple linear regression model constructed here consists of β0 representing the intercept, while β1 to β5 correspond to the coefficients of the variables, and ϵ denotes the error term. Analysis of these coefficients reveals the relationship between w_new and the metrics as follows: β0, the intercept, at 0.7902 indicates the predicted value of w_new when all variables are at zero. β1 (R 2 -training) suggests a decrement of 0.1588 in w_new per unit increase in R 2 -training, holding other variables constant. Conversely, β2 (R 2 -validation) shows an increase of 0.4065 in w_new per unit rise in R 2 -validation, with other variables fixed. β3 (MAE) implies w_new increases by 0.6306 for each unit escalation in MAE, controlling for other variables. β4 (RMSE) indicates a reduction of 1.5866 in w_new per unit augmentation in RMSE, maintaining other variables. β5 (MSE) reveals an increase of 0.1938 in w_new for each unit increase in MSE, with other variables steady. The error term (ϵ) coefficient demonstrates a marginal positive influence on w_new, quantified at 0.0002. Statistical significance was assessed using associated p-values, where p-values < 0.05 were considered significant. The analysis indicates significant coefficients for R 2 -training, R 2 -validation, RMSE, and the error term, while MAE and MSE may not be statistically significant predictors of w_new when considered alongside other variables. Overall, R 2 -validation and RMSE emerge as the most influential factors impacting w_new, based on their coefficient magnitudes and statistical significance levels. These findings suggest that factors such as PCA/features components, parameters of each model, and cross-validation times have less impact on w_new.

The effects of different factors on w_new in individual models

The exploration of various machine learning models unveiled consistent patterns in the relationship between the parameter w_new and model performance metrics. Across models like Adaboost, Decision Tree, Elastic Net Regression, SVR, and KNN, w_new displayed discernible associations. Notably, positive correlations were observed between w_new and certain performance indicators like 'Cross-validation' and R 2 -validation, suggesting that higher w_new values align with improved validation scores. Conversely, w_new consistently exhibited negative relationships with error metrics such as RMSE, MAE, and MSE, indicating that an increase in w_new corresponded to decreased error rates across models. Additionally, some models showcased nuanced relationships between w_new and specific parameters, like 'Minimum sample split' in the Decision Tree and 'Model gamma' in SVR. Overall, the consistent trends suggest that w_new plays a significant role in influencing model performance, particularly in relation to validation scores and error metrics, across diverse machine learning models [ 48 ]. See the Supplementary Fig. 1 for more details.

The effects of hyperparameters on w_new in individual models

In computational modeling, the relationship between model complexity and hyperparameters, particularly in KNN models, highlights the critical influence of the number of neighbors (“K”) on model performance, showing a negative correlation of -0.877 with w_new. Decreasing “K” simplifies the model and improves prediction accuracy, notably in shape similarity and pharmacophore models, diverging from other QSAR model outcomes [ 49 ]. For Elastic Net models, model_alpha and the “L1 Ratio” hyperparameters significantly impact complexity, with negative correlations of − 0.349 and − 0.978 with w_new, respectively, indicating their strong influence on reducing model complexity [ 50 ]. Refer to Fig.  6 for a visualization of these relationships.

Bar plot illustrating the correlation strengths between various model parameters and the metric ‘w_new’ across different machine learning models. Each bar represents the correlation value of a specific parameter with ‘w_new’ for a given model. Positive values indicate a direct relationship, while negative values suggest an inverse relationship between the parameter and ‘w_new’

The Random Forest model demonstrates complexity modulation through hyperparameters, where “Max depth” and “Number of Estimators” exhibit high positive correlations with w_new, indicating an increase in model intricacy as these parameters increase [ 51 ] as depicted in Fig.  6 . Conversely, “Min sample leaf” and “Min samples split” show significantly high negative correlations with w_new, implying a decrease in w_new with the escalation of these parameters [ 52 ]. In the Adaboost models, the “Number of Estimators” shows a slight positive correlation (0.014) with w_new, while the “learning rate” exhibits a significant negative correlation (− 0.321), suggesting a decrease in model complexity with a higher learning rate. In SVR, the “Model cost” and “Model gamma” parameters show negative correlations of − 0.247 and − 0.149 with w_new, respectively, indicating their roles in slightly reducing model complexity as they increase [ 53 , 54 ].

Overall, the analysis highlights the varied impacts of hyperparameters on model complexity, with some leading to increased complexity and others to simplification, depending on the model and hyperparameter [ 55 ]. Simplified models favored in this study enhance interpretability and computational efficiency, offering advantages in real-time scenarios and environments with limited computing capacity [ 56 ]. Moreover, their simplicity is advantageous in situations with restricted data availability, showcasing superior performance relative to more complex models prone to overfitting and sensitivity to noise in sparse datasets [ 57 ].

Enrichment metrics for the consensus holistic scoring in comparison to individual screening methods

In evaluating various screening methods against consensus screening for different protein targets, we detailed their performance metrics, including AUC ROC, EF1%, EF5%, decoy percentage at 1%, and Boltzmann-Enhanced Discrimination of ROC (BEDROC) values, as defined in the Supplementary information and Fig.  7 . For the AKT1 protein target, docking screening exhibited superior performance with an AUC ROC score of 0.87, marginally higher than the consensus score of 0.85. Similarity screening followed with a score of 0.79, while Pharmacophore and QSAR methods registered scores of 0.74 and 0.64, respectively. In terms of EF1%, Similarity screening outperformed with a score of 63.0, surpassing the consensus score of 57.5. Docking and QSAR methods both achieved 40.0, and Pharmacophore screening was lower at 22.68. BEDROC scores showed Similarity screening leading with 0.5443, above the consensus of 0.523, followed by QSAR (0.3935), Docking (0.3174), and Pharmacophore (0.224).

Area under the ROC curve for ( A ) consensus scoring method in protein targets involved in this study in comparison to ( B ) each target evaluated by four different screening methods; QSAR (PIC 50 ), Docking, pharmacophore, and shape similarity screenings in comparison with the consensus scoring

For the CDK2 protein, docking screening again excelled with an AUC score of 0.84, slightly above the consensus of 0.83. Similarity and Pharmacophore screenings scored 0.61 and 0.59, respectively, with QSAR trailing at 0.56. EF1% values showed Docking leading significantly with 78.12, well above the consensus of 65.0. QSAR recorded 45.36, with Similarity and Pharmacophore screenings at 25.2 and 27.72, respectively. BEDROC values for Docking and QSAR were close to the consensus score of 0.4192, at 0.4864 and 0.3203, respectively, while Similarity and Pharmacophore screenings had lower values of 0.2168 and 0.2354. This comprehensive evaluation, detailed in Supplementary Table 2, underscores the variable efficacy of screening methods across protein targets, informing their strategic application in virtual screening.

In the evaluation of DPP4 using consensus scoring, the QSAR screening method's AUC score of 0.82 is closely matched to the consensus of 0.84, with Pharmacophore and Similarity methods yielding lower scores of 0.65 and 0.66, respectively, and docking the lowest at 0.56. For EF1%, QSAR and consensus both achieve 46.81, with Similarity at 36.17, Pharmacophore at 31.91 and docking significantly lower at 8.51. In BEDROC scores, QSAR exceeds consensus with 0.4893 versus 0.4559, followed by Pharmacophore and Similarity methods at 0.381 and 0.3646, respectively, and docking considerably behind at 0.0969.

For the EGFR protein, Pharmacophore screening excels with an AUC of 0.93, exceeding the consensus of 0.77. Similarity screening is close to consensus at 0.73, with QSAR at 0.64, and docking significantly behind at 0.36. QSAR's EF1% of 30.3 is near the consensus of 34.67, with Similarity and Docking trailing at 13.86 and 14.18, respectively, and Pharmacophore notably lower at 3.96. BEDROC metrics show all methods aligning closely around the consensus of 0.6139, except for QSAR which lags at 0.3649. Refer to Fig.  7 for ROC curves of the various scoring and screening methodologies.

For the AA2AR, the QSAR screening method achieved an AUC of 0.78, marginally higher than the consensus of 0.77, followed by Docking at 0.72. Pharmacophore screening recorded a lower AUC of 0.54, with Similarity trailing at 0.4. In the EF1% evaluation, Pharmacophore led with 50.4, above the consensus of 45.36, and docking at 42.84, while QSAR and Similarity both reported 0.0. BEDROC scores for Pharmacophore and Docking were close to the consensus of 0.4401, at 0.3962 and 0.3974, respectively.

In contrast, for the p53 protein, Pharmacophore screening achieved the highest AUC of 0.93, slightly above the consensus of 0.90, with Docking at 0.77 and Similarity at 0.64. QSAR was notably lower at 0.49. Pharmacophore screening exhibited outstanding EF1% performance at 88.96, surpassing the consensus of 76.82. In BEDROC metrics, Pharmacophore again led with 0.4661, exceeding the consensus of 0.4336, followed by Docking and Similarity at 0.3553 and 0.2952, respectively, and QSAR at 0.1445.

In the case of the PPARG protein, the Pharmacophore screening method achieved an AUC ROC of 0.80, near the consensus of 0.90, with Similarity, QSAR, and Docking methods following at 0.69, 0.67, and 0.66, respectively. In EF1%, Docking led with 48.67, exceeding the consensus of 42.35. Docking also topped the BEDROC metric with 0.3135, surpassing the consensus of 0.2896, with Similarity and QSAR at 0.1354 and 0.2372, respectively. Regarding the TDP1 protein, Pharmacophore screening outperformed with an AUC of 0.84, above the consensus of 0.73. Similarity matched the consensus at 0.73, while Docking and QSAR lagged with 0.4 and 0.3, respectively. For BEDROC, Pharmacophore significantly led with 0.2319, doubling the consensus of 0.1184, with Similarity and Docking at 0.1271 and 0.0623, and QSAR at 0.0163, indicating a marked disparity in the early detection of actives across screening methods.

The consensus holistic scoring in comparison to other consensus virtual screening methods

A comparative analysis of three consensus docking approaches reveals distinct advantages and disadvantages. Houston and Walkinshaw [ 6 ] demonstrated improved pose prediction accuracy (82% success rate) and reduced false positives by integrating multiple docking programs, albeit with increased computational costs and potential rise in false negatives. Besides, Ochoa, Palacio-Rodriguez [ 10 ] introduced a score-based consensus docking approach with higher success rates in pose prediction and consideration of biological target flexibility, but its efficacy may depend on individual docking program performance and could introduce biases toward certain molecules or poses. The pose rank consensus (PRC) method [ 11 ], significantly improves systematic performance and hit rates at minimal computational cost, yet its effectiveness relies on individual docking program performance and may have limitations in scenarios with few ligands or underperforming target proteins. Studies indicate that increasing time allocated for consensus docking calculations may not significantly improve method performance, highlighting nuanced trade-offs between accuracy, computational efficiency, and inherent limitations of consensus docking in virtual screening [ 58 ].

The combined use of ligand- and structure-based methodologies in computer-aided drug design optimizes chemical and biological data integration, enhancing efficacy through synergistic exploitation of their respective advantages while mitigating individual drawbacks. This integrated approach typically outperforms standalone methods, especially when employing parallel or other integrated techniques to automate and streamline virtual screening processes [ 15 ]. However, challenges persist, including the subjective and intricate nature of sequential approach selection, the complexity of method combination in parallel strategies, and limitations in accurately predicting future virtual screening performance through retrospective analyses. Prospective assessments, though more indicative of method efficacy in identifying diverse new hits, demand significantly greater resources and expertise for execution [ 59 ].

Swann, Brown [ 17 ] devised a novel consensus method merging structure-based and ligand-based screening into a unified probabilistic framework, demonstrating superior performance compared to individual metrics. This approach integrates comprehensive chemical and structural data, enhancing the diversity of identified active compounds and offering a fresh perspective on chemical similarity for drug discovery. Despite its transformative potential in virtual screening, challenges arise from the complexity of developing and validating Probability Assignment Curves (PACs), potentially restricting accessibility to researchers without computational expertise. Furthermore, the method’s efficacy depends on data quality, necessitating caution regarding generalizability and advocating for inclusive tools or guidelines to improve accessibility. Extensive validation efforts underscore concerns regarding dataset biases, highlighting the need for broader validation to ensure method robustness and mitigate overfitting risks.

The consensus holistic scoring method showcased in this study outperforms singular methodologies in identifying potential hit compounds across diverse protein targets. Introduction of the “w_new” metric enhances drug discovery efficacy by refining ML model rankings, albeit without consistently yielding optimal ROC curves. Nevertheless, it effectively prioritizes compounds with higher experimental activity, ensuring a robust screening process. Validation against biases between active compounds and decoys enhances prediction reliability. However, the method primarily serves as a scoring tool for refining true positives and does not offer insights into binding pose predictions. Integration of multiple screening methods and ML models demands substantial computational resources and expertise, along with labor and time-intensive validation and tuning for each target-specific ML model.

Combining diverse methodologies in drug discovery yields comprehensive insights into ligand-receptor interactions, crucial for designing potent binders. Molecular docking predicts binding affinity and ligand orientation in proteins, unveiling interaction insights. Pharmacophore modeling identifies critical features for spacial arrangement required for binding, guiding enhanced compound design. 3D-QSAR analysis quantitatively links ligand structure to biological activity, enabling activity predictions for new compounds [ 60 ]. Furthermore, the value of molecular similarity in drug discovery becomes apparent when integrating 2D and 3D shape similarity methods, which contribute significantly to a more comprehensive workflow for identifying molecules with similar structures and properties [ 61 ]. Integrating these methods offers a holistic view, elucidating key structural elements and their impact on activity. This integrated approach ensures precise predictions, empowering rational design and optimization of novel drug candidates.

Based on our analysis, the incorporation of weighted machine learning algorithms streamlined the identification of the optimal model among the twelve machine learning models introduced in this study, which encompass commonly-utilized ML models. This coding framework holds applicability across a wide spectrum of applications and can readily integrate the novel “w_new” formula into various contexts, particularly within continuous regression models, whether applied to virtual screening or other domains. The amalgamation of three key performance enhancers, namely error reduction, R 2 enhancement across training and validation sets, and mitigation of overfitting risks by minimizing the disparity between R 2 values in training and validation, represents, to the best of our knowledge, a novel conceptual advance.

In this investigation, we devised a streamlined approach for the examination of active and decoy distribution in the datasets, intending to identify bias and accurately evaluate the performance metrics of models. A three-stage workflow was developed for dataset validation, including quantification of physicochemical properties, diversity analysis through fragment fingerprints, and the graphical depiction of compound distributions using 2D PCA. This methodology not only addressed biases from uneven physicochemical property distributions and analogue bias but also illustrated structural diversity. The results, supported by comparisons with Maximum Unbiased Validation (MUV) datasets, indicated a high degree of similarity in distribution patterns, except for specific datasets with unique compositions. The diversity analysis further underscored the methodological strength, showing a balanced chemical class diversity and an insightful disparity in diversity ranks towards actives. This comprehensive approach, marked by a meticulous assessment of physicochemical properties and innovative use of similarity mapping and PCA, contributed to a more precise evaluation of the chemical space and dataset biases.

The study explores the factors impacting w_new and how model complexity interacts with performance metrics. Correlation analyses reveal positive correlations between w_new and R 2 -training and R 2 -validation, while error metrics like MAE, RMSE, and MSE negatively correlate with w_new. Multiple linear regression reveals that among the considered variables, R 2 -validation and RMSE most significantly affect w_new. Overall, hyperparameters can either increase or decrease model complexity depending on the specific model and parameter. Besides, the models in this study consistently favor simplicity, which enhances interpretability, computational efficiency, and robustness in data-scarce scenarios, making them suitable for diverse applications.

Across all models, the average external validation R 2 value is ~ 0.724, indicating a moderate to high performance with a standard deviation of 0.088, highlighting significant variability across models. The R 2 values range from 0.586 to 0.905. The GPCR protein AA2AR, using the pharmacophore scoring method with the 'KNN' machine learning model, achieved the highest external validation R 2 of 0.905, demonstrating excellent predictivity with R 2 -train of 0.999 and R 2 -val of 0.88. In contrast, the protein p53, utilizing the 2D fingerprint shape similarity method with the ‘Adaboost’ model, showed the lowest R 2 -ext of 0.586, despite a significantly high R 2 -train of 0.969 and R 2 -val of 0.691, suggesting limitations in generalizability possibly due to dataset specifics, overfitting, or inherent protein characteristics.

In the context of the enrichment studies, it is of note that the area under the ROC curve achieved via consensus screening within the framework of the AA2AR receptor exhibits a performance level closely comparable to that of the QSAR screening, as expounded upon in the previous section. This modest augmentation in the ROC curve’s AUC assumes negligible significance when we ascertain that the initial active compound in the dataset, CHEMBL1093479, attains prioritization after an extensive cohort comprising 91 decoy compounds within QSAR screening. Meanwhile, in the consensus scoring, the first seven positions are occupied by active compounds, manifesting potency levels extending up to a PIC 50 value of 10. This observation receives additional corroboration through the inclusion of enrichment metrics delineated within Supplementary Table 2. These metrics encompass the BEDROC, along with the percentages denoting the early fractions (EF1% and EF5%), as well as the fraction of decoys at the 1% threshold.

A parallel scenario unfolds in our evaluation of the EGFR protein target. In the domain of consensus scoring, the top four compounds are identified as active against EGFR, exhibiting PIC 50 values ranging from 6.77 to 9.25. In contrast, the pharmacophore screening for EGFR, yielding a notably higher ROC value of 0.93, positions the first active compound, CHEMBL451513, and the second active compound, CHEMBL516022, at significantly lower ranks within the entire compound pool in the enrichment study, specifically at the 43rd and 47th positions, respectively. Remarkably, among the top-ranked compounds prominently enriched in the top ten ranks in the consensus scoring results for EGFR, are compounds such as CHEMBL63786, CHEMBL176582, and CHEMBL460723, each exhibiting the highest PIC 50 values within the dataset, measuring 11.52, 11, and 9.25, respectively.

Continuing within the same analytical framework, we assess the ROC AUC for the AKT1 target when comparing consensus scoring to docking screening. While the AUC values appear to exhibit minimal disparity, a more discerning examination reveals that the metrics of EF1% and BEDROC unequivocally favor the consensus scoring approach. Furthermore, when we consider additional metrics such as EF5% and the decoy percentage at the initial 1%, it becomes evident that shape similarity screening outperforms the docking method in this context. It is crucial to emphasize that a singular performance metric cannot definitively establish the superiority of one scoring or screening method over another. Hence, a comprehensive evaluation must also consider the prioritization of active compounds within each method. In Fig.  8 , we observe that compounds identified as top-ranked by the consensus scoring method exhibit superior PIC 50 values compared to those identified by the docking approach. Specifically, CHEMBL212566 and CHEMBL1098938, top-ranked by consensus scoring, display PIC 50 values of 8.49 and 9.70, respectively. In the same vein, the consensus scoring prominently enriches CHEMBL523586 at the 24th rank. However, within the docking approach, despite its noteworthy PIC 50 value of 10.52, CHEMBL523586 assumes a considerably lower rank, standing at 1816th. Similarly, in the shape similarity screening, its ranking descends even further, settling at the 5037th position, thereby unveiling a substantial divergence across these methodologies. These findings underscore the multifaceted nature of our evaluation, where a holistic assessment considers not only quantitative metrics but also the prioritization of active compounds as a pivotal aspect of the screening process.

Top-ranked compounds in AKT1 and CDK2 targets in consensus and docking methodologies with their respective PIC 50 values

In the CDK2 screening analysis, it is notable that the enrichment metrics derived from consensus and docking screening exhibited a close alignment concerning various parameters, including AUC, EF1%, EF5%, BEDROC, and Decoy Percentage at 1%, albeit with a slight advantage observed in favor of the docking method. However, a more nuanced assessment reveals that the consensus scoring approach excelled in the prioritization of compounds with higher PIC 50 values. This distinction is particularly evident when scrutinizing Fig.  8 , which highlights the top four active compounds with PIC 50 values ranging from 6.60 to 8.05 for consensus scoring, as opposed to a narrower range of 6.33 to 7.48 for the docking screening. Furthermore, it is noteworthy that compounds possessing the highest PIC 50 values within the CDK2 dataset received more favorable rankings within the consensus scoring methodology compared to the docking screening. For instance, the compound with the highest PIC 50 value, namely CHEMBL360520, attaining 9.52, was positioned at the 18th rank in the consensus scoring, while the docking method placed it considerably lower at the 3415th position. Similarly, the second top-ranked compound in terms of PIC 50 (CHEMBL261720) within the dataset achieved a ranking of 28th in consensus scoring, while the docking method assigned it a lower ranking of 44th.

In an alternative context, the performance of consensus scoring for TDP1 demonstrated diminished robustness when compared to its efficacy in assessing other macromolecules. Notably, pharmacophore screening exhibited markedly superior performance across all evaluation metrics in contrast to the consensus screening approach. This distinctive behavior observed for the TDP1 target can be ascribed to the limited activity range present within the datasets. Intriguingly, the consensus scoring for TDP1, conducted using commercially available software as described by Moshawih, Goh [ 22 ], yielded a remarkably high AUC ROC value of ~ 0.98. This exceptional outcome can be attributed to meticulous process optimization, including the selection of an optimal model and a well-suited set of features. Additionally, in this study, the decoy pool consisted of 2700 compounds for the same dataset, introducing an added layer of complexity to the analysis. In a different context, it is noteworthy that the p53 dataset is relatively small, consisting of only 20 active compounds (and 5 external validation datasets) primarily comprising anthraquinones and chalcones. Nevertheless, the consensus methodology demonstrated exemplary performance across all enrichment metrics, mirroring the trends observed with the pharmacophore approach. Moreover, the consensus scoring for p53 was also performed using commercial software in a separate study (data not published), and the resulting AUC ROC and other pertinent metrics closely paralleled the findings reported herein, with a value of 0.90. This observation suggests that consensus scoring has the capacity to effectively identify optimal characteristics from diverse screening methodologies across a wide range of scenarios and combine them to obtain the best enrichment in virtual screening.

In this investigation, we undertook a comprehensive analysis involving eight diverse protein targets across various functional categories. Our primary objective was to evaluate the efficacy of a consensus holistic virtual screening approach across heterogeneous datasets. Significantly, while the PIC 50 values for some protein targets displayed a constrained distribution, emphasizing the limited range of activities, the shape similarity scores followed by other screenings exhibited consistent and widespread patterns across all targets. Particularly, when combined with all of the screening methods through a consensus approach, it is expected to emerge as a potent strategy, demonstrating that consensus scoring selects the most favorable aspects from multiple screening metrics.

This investigation integrated a novel methodology for analyzing active and decoy distribution biases in datasets, which significantly impacted model performance and highlighted the importance of dataset validation in virtual screening. Our quest for a robust consensus scoring methodology for a holistic virtual screening led us to employ a variety of machine learning models devised with a novel formula that amalgamates five performance metrics into a unified measure called w_new. The greater weight assigned (w_new) signifies a robust model performance, characterized by higher R 2 -training and -validation scores, reduced MAE, RMSE, and MSE values, and minimized disparity between R 2 -train and -validation and vice versa. This comprehensive study unveiled a spectrum of performance metrics among different models employed.

In our endeavor to elucidate the factors influencing w_new values and assess the impact of model complexity on performance metrics, we conducted an exhaustive analysis. Our investigation revealed that R 2 -validation and RMSE are pivotal factors influencing w_new, exhibiting positive and negative correlations, respectively. These findings underscore models used in this study consistently prioritize simplicity, leading to improved computational efficiency, data efficiency, and practical applicability. Furthermore, our study shed light on the nuanced relationships between w_new and various model-specific parameters, providing insights into the interplay between model complexity and performance metrics.

Overall, weighted machine learning models find utility across diverse domains and are not restricted to virtual screening, where the primary objective is the identification of optimal, high-performing, and resilient models. Besides, this comprehensive analysis underscores the importance of considering not only quantitative metrics but also the prioritization of active compounds, which can vary significantly across different methods when choosing screening and scoring methodologies. This analysis emphasizes the effectiveness of consensus scoring as a crucial virtual screening technique, often yielding superior performance in terms of AUC, early detection of actives, prioritizing compounds with the highest biological activities, or a combination of these factors. These findings contribute significantly to advancing our understanding of screening techniques' performance in diverse protein target contexts, ultimately enhancing the effectiveness of virtual screening approaches.

Availability of data and materials

The code snippets developed and other code implementations utilized in this study can be accessed via the following GitHub link: https://github.com/Saeedmomo/Consensus_Holistic_Virtual_Screening.git . Furthermore, the datasets used in this research are also accessible within the same GitHub repository.

Lionta E, Spyrou G, Vassilatis KD, Cournia Z (2014) Structure-based virtual screening for drug discovery: principles, applications and recent advances. Curr Top Med Chem 14(16):1923–1938

Article   CAS   PubMed   PubMed Central   Google Scholar  

Moshawih S, Goh HP, Kifli N, Idris AC, Yassin H, Kotra V et al (2022) Synergy between machine learning and natural products cheminformatics: application to the lead discovery of anthraquinone derivatives. Chem Biol Drug Design. 100:185–217

Article   CAS   Google Scholar  

Baber JC, Shirley WA, Gao Y, Feher M (2006) The use of consensus scoring in ligand-based virtual screening. J Chem Inf Model 46(1):277–288

Article   CAS   PubMed   Google Scholar  

Stanton DT (1999) Evaluation and use of BCUT descriptors in QSAR and QSPR studies. J Chem Inf Comput Sci 39(1):11–20

Pirard B, Pickett SD (2000) Classification of kinase inhibitors using BCUT descriptors. J Chem Inf Comput Sci 40(6):1431–1440

Houston DR, Walkinshaw MD (2013) Consensus docking: improving the reliability of docking in a virtual screening context. J Chem Inf Model 53(2):384–390

Huey R, Morris G. AutoDock tools. La Jolla, CA, USA: The Scripps Research Institute. 2003.

Allen WJ, Balius TE, Mukherjee S, Brozell SR, Moustakas DT, Lang PT et al (2015) DOCK 6: impact of new features and current docking performance. J Comput Chem 36(15):1132–1156

Trott O, Olson AJ (2010) AutoDock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. J Comput Chem 31(2):455–461

Ochoa R, Palacio-Rodriguez K, Clemente CM, Adler NS (2021) dockECR: open consensus docking and ranking protocol for virtual screening of small molecules. J Mol Graph Model 109:108023

Scardino V, Bollini M, Cavasotto CN (2021) Combination of pose and rank consensus in docking-based virtual screening: the best of both worlds. RSC Adv 11(56):35383–35391

Yang J-M, Chen Y-F, Shen T-W, Kristal BS, Hsu DF (2005) Consensus scoring criteria for improving enrichment in virtual screening. J Chem Inf Model 45(4):1134–1146

Stahura FL, Bajorath J (2004) Virtual screening methods that complement HTS. Comb Chem High Throughput Screen 7(4):259–269

Tanrikulu Y, Krüger B, Proschak E (2013) The holistic integration of virtual screening in drug discovery. Drug Discov Today 18(7–8):358–364

Article   PubMed   Google Scholar  

Drwal MN, Griffith R (2013) Combination of ligand- and structure-based methods in virtual screening. Drug Discov Today Technol 10(3):e395–e401

McInnes C (2007) Virtual screening strategies in drug discovery. Curr Opin Chem Biol 11(5):494–502

Swann SL, Brown SP, Muchmore SW, Patel H, Merta P, Locklear J et al (2011) A unified, probabilistic framework for structure-and ligand-based virtual screening. J Med Chem 54(5):1223–1232

Ericksen SS, Wu H, Zhang H, Michael LA, Newton MA, Hoffmann FM et al (2017) Machine learning consensus scoring improves performance across targets in structure-based virtual screening. J Chem Inf Model 57(7):1579–1590

Gentile F, Yaacoub JC, Gleave J, Fernandez M, Ton A-T, Ban F et al (2022) Artificial intelligence-enabled virtual screening of ultra-large chemical libraries with deep docking. Nat Protoc 17(3):672–697

Ton AT, Gentile F, Hsing M, Ban F, Cherkasov A (2020) Rapid identification of potential inhibitors of SARS-CoV-2 main protease by deep docking of 1.3 billion compounds. Mol Informat. 39(8):2000028

Yaacoub JC, Gleave J, Gentile F, Stern A, Cherkasov A (2022) DD-GUI: a graphical user interface for deep learning-accelerated virtual screening of large chemical libraries (Deep Docking). Bioinformatics 38(4):1146–1148

Moshawih S, Goh HP, Kifli N, Darwesh MAE, Ardianto C, Goh KW, et al. Identification and optimization of TDP1 inhibitors from anthraquinone and chalcone derivatives: consensus scoring virtual screening and molecular simulations. J Biomol Struct Dynam. 2023:1–25.

Kim S, Thiessen PA, Bolton EE, Chen J, Fu G, Gindulyte A et al (2015) PubChem substance and compound databases. Nucleic Acids Res 44(D1):D1202–D1213

Article   PubMed   PubMed Central   Google Scholar  

Mysinger MM, Carchia M, Irwin JJ, Shoichet BK (2012) Directory of useful decoys, enhanced (DUD-E): better ligands and decoys for better benchmarking. J Med Chem 55(14):6582–6594

Adeshina YO, Deeds EJ, Karanicolas J (2020) Machine learning classification can reduce false positives in structure-based virtual screening. Proc Natl Acad Sci 117(31):18477–18488

Feng M, Heinzelmann G, Gilson MK (2022) Absolute binding free energy calculations improve enrichment of actives in virtual compound screening. Sci Rep 12(1):13640

Vieira TF, Sousa SF (2019) Comparing AutoDock and Vina in ligand/decoy discrimination for virtual screening. Appl Sci 9(21):4538

Sieg J, Flachsenberg F, Rarey M (2019) In need of bias control: evaluating chemical data for machine learning in structure-based virtual screening. J Chem Inf Model 59(3):947–961

Li Y, Yang J (2017) Structural and sequence similarity makes a significant impact on machine-learning-based scoring functions for protein–ligand interactions. J Chem Inf Model 57(4):1007–1012

Riniker S, Landrum GA (2013) Open-source platform to benchmark fingerprints for ligand-based virtual screening. J Cheminformat 5(1):26

Rohrer SG, Baumann K (2009) Maximum unbiased validation (MUV) data sets for virtual screening based on PubChem bioactivity data. J Chem Inf Model 49(2):169–184

RDKit. Open-source cheminformatics. Available from: https://www.rdkit.org

Moshawih S, Hadikhani P, Fatima A, Goh HP, Kifli N, Kotra V et al (2022) Comparative analysis of an anthraquinone and chalcone derivatives-based virtual combinatorial library. A cheminformatics “proof-of-concept” study. J Mol Graph Modelling. 117:108307

Moshawih S, Lim AF, Ardianto C, Goh KW, Kifli N, Goh HP et al (2022) Target-based small molecule drug discovery for colorectal cancer: a review of molecular pathways and in silico studies. Biomolecules 12(7):878

Chua HM, Moshawih S, Goh HP, Ming LC, Kifli N (2023) Insights into the computer-aided drug design and discovery based on anthraquinone scaffold for cancer treatment: a protocol for systematic review. PLoS ONE 18(9):e0290948

Berman HM, Battistuz T, Bhat TN, Bluhm WF, Bourne PE, Burkhardt K et al (2002) The protein data bank. Acta Crystallogr D Biol Crystallogr 58(6):899–907

Schrodinger. Schrodinger 2022-3. 2022.

Chen D, Zheng J, Wei G-W, Pan F (2021) Extracting predictive representations from hundreds of millions of molecules. J Phys Chem Lett 12(44):10793–10801

Cohen J, Cohen P, West SG, Aiken LS (2013) Applied multiple regression/correlation analysis for the behavioral sciences. Routledge, USA

Book   Google Scholar  

Chai T, Draxler RR (2014) Root mean square error (RMSE) or mean absolute error (MAE)?—arguments against avoiding RMSE in the literature. Geoscientif Model Develop 7(3):1247–1250

Article   Google Scholar  

García S, Luengo J, Herrera F (2015) Data normalization. In: Data preprocessing in data mining. Springer, Cham

James G, Witten D, Hastie T, Tibshirani R (2013) An introduction to statistical learning. Springer, Cham

Sander T, Freyss J, von Korff M, Rufener C (2015) DataWarrior: an open-source program for chemistry aware data visualization and analysis. J Chem Inf Model 55(2):460–473

Scantlebury J, Vost L, Carbery A, Hadfield TE, Turnbull OM, Brown N et al (2023) A small step toward generalizability: training a machine learning scoring function for structure-based virtual screening. J Chem Inf Model 63(10):2960–2974

Yeolekar A, Patel S, Talla S, Puthucode KR, Ahmadzadeh A, Sadykov VM, et al., editors. Feature selection on a flare forecasting testbed: a comparative study of 24 methods. In: 2021 International Conference on Data Mining Workshops (ICDMW); 2021 7–10 Dec. 2021

Andersson PL, Fick J, Rännar S (2011) A multivariate chemical similarity approach to search for drugs of potential environmental concern. J Chem Inf Model 51(8):1788–1794

Svetnik V, Wang T, Tong C, Liaw A, Sheridan RP, Song Q (2005) Boosting: an ensemble learning tool for compound classification and QSAR modeling. J Chem Inf Model 45(3):786–799

Géron A. Hands-on machine learning with Scikit-Learn, Keras, and TensorFlow: O’Reilly Media, Inc.; 2019.

Itskowitz P, Tropsha A (2005) k nearest neighbors QSAR modeling as a variational problem: theory and applications. J Chem Inf Model 45(3):777–785

Matsui H, Konishi S (2011) Variable selection for functional regression models via the L1 regularization. Comput Stat Data Anal 55(12):3304–3310

Jogdeo AA, Patange AD, Atnurkar AM, Sonar PR (2023) Robustification of the random forest: a multitude of decision trees for fault diagnosis of face milling cutter through measurement of spindle vibrations. J Vib Eng Technol 12:1–19

Google Scholar  

Yajima H, Derot J (2018) Application of the Random Forest model for chlorophyll- a forecasts in fresh and brackish water bodies in Japan, using multivariate long-term databases. J Hydroinf 20(1):206–220

Goodfellow I, Bengio Y (2016) Deep learning. MIT Press, USA

Bergstra J, Bengio Y (2012) Random search for hyper-parameter optimization. J Mach Learn Res 13(2)

Hansen KB (2020) The virtue of simplicity: on machine learning models in algorithmic trading. Big Data Soc 7(1):2053951720926558

Kumar P, Sinha K, Nere NK, Shin Y, Ho R, Mlinar LB et al (2020) A machine learning framework for computationally expensive transient models. Sci Rep 10(1):11492

Sarker IH (2021) Machine learning: algorithms, real-world applications and research directions. SN Comput Sci 2(3):160

Kukol A (2011) Consensus virtual screening approaches to predict protein ligands. Eur J Med Chem 46(9):4661–4664

Varela-Rial A, Majewski M, De Fabritiis G (2022) Structure based virtual screening: fast and slow. Wiley Interdiscip Rev Comput Mol Sci 12(2):e1544

Lu S, Liu H-C, Chen Y-D, Yuan H-L, Sun S-L, Gao Y-P et al (2011) Combined pharmacophore modeling, docking, and 3D-QSAR studies of PLK1 inhibitors. Int J Mol Sci 12(12):8713–8739

Kumar A, Zhang KYJ (2018) Advances in the development of shape similarity methods and their application in drug discovery. Front Chem 6

Download references

All expenses associated with this paper, including the publication fee, were generously covered by the Council for Research and Advancement in Technology and Science (CREATES) MTIC/CREATES PROJECT UBD 006, under the Ministry of Transport and Infocommunications (MTIC), specifically through its Technology and Innovation Division, in Negara Brunei Darussalam.

Author information

Authors and affiliations.

PAPRSB Institute of Health Sciences, Universiti Brunei Darussalam, Gadong, Brunei Darussalam

Said Moshawih, Hui Poh Goh, Nurolaini Kifli & Long Chiau Ming

Faculty of Computing and Engineering, Quest International University, Ipoh, Malaysia

Zhen Hui Bu & Lam Hong Lee

Faculty of Data Science and Information Technology, INTI International University, Nilai, Malaysia

Said Moshawih & Khang Wen Goh

School of Medical and Life Sciences, Sunway University, Sunway City, Malaysia

Long Chiau Ming

You can also search for this author in PubMed   Google Scholar

Contributions

Conceptualization: S.M.; Data curation: Z.H.B, K.W.G., H.P.G.; Formal Analysis: S.M., L.C.M., N.K.; Funding acquisition: L.C.M., L.H.L., K.W.G.; Methodology: S.M., Z.H.B., L.H.L.; Project administration: S.M., L.C.M.; Resources: N.K., H.P.G.; Software: S.M., L.H.L.; Supervision: L.C.M., L.H.L.; Validation: S.M., Z.H.B.; Visualization: S.M.; Writing—original draft: S.M.; Writing—review and editing: L.C.M., L.H.L., K.W.G.

Corresponding author

Correspondence to Said Moshawih .

Ethics declarations

Ethics approval and consent to participate.

No ethics approval or human participants were included in this study.

Consent for publication

No human participants were included in this study.

Competing interests

The authors declare no competing interests.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Supplementary material 1., supplementary material 2., supplementary material 3., rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Moshawih, S., Bu, Z.H., Goh, H.P. et al. Consensus holistic virtual screening for drug discovery: a novel machine learning model approach. J Cheminform 16 , 62 (2024). https://doi.org/10.1186/s13321-024-00855-8

Download citation

Received : 03 December 2023

Accepted : 10 May 2024

Published : 28 May 2024

DOI : https://doi.org/10.1186/s13321-024-00855-8

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Consensus scoring
• Virtual screening
• Machine learning models
• Pharmacophore
• Shape similarity

Journal of Cheminformatics

ISSN: 1758-2946

• Submission enquiries: [email protected]

Information

• Author Services

Initiatives

You are accessing a machine-readable page. In order to be human-readable, please install an RSS reader.

All articles published by MDPI are made immediately available worldwide under an open access license. No special permission is required to reuse all or part of the article published by MDPI, including figures and tables. For articles published under an open access Creative Common CC BY license, any part of the article may be reused without permission provided that the original article is clearly cited. For more information, please refer to https://www.mdpi.com/openaccess .

Feature papers represent the most advanced research with significant potential for high impact in the field. A Feature Paper should be a substantial original Article that involves several techniques or approaches, provides an outlook for future research directions and describes possible research applications.

Feature papers are submitted upon individual invitation or recommendation by the scientific editors and must receive positive feedback from the reviewers.

Editor’s Choice articles are based on recommendations by the scientific editors of MDPI journals from around the world. Editors select a small number of articles recently published in the journal that they believe will be particularly interesting to readers, or important in the respective research area. The aim is to provide a snapshot of some of the most exciting work published in the various research areas of the journal.

Original Submission Date Received: .

• Active Journals
• Find a Journal
• Proceedings Series
• For Authors
• For Reviewers
• For Editors
• For Librarians
• For Publishers
• For Societies
• For Conference Organizers
• Open Access Policy
• Institutional Open Access Program
• Special Issues Guidelines
• Editorial Process
• Research and Publication Ethics
• Article Processing Charges
• Testimonials
• Preprints.org
• SciProfiles
• Encyclopedia

Article Menu

• Subscribe SciFeed
• Recommended Articles
• Author Biographies
• PubMed/Medline
• Google Scholar
• on Google Scholar
• Table of Contents

Find support for a specific problem in the support section of our website.

Please let us know what you think of our products and services.

Visit our dedicated information section to learn more about MDPI.

JSmol Viewer

Microfluidics in high-throughput drug screening: organ-on-a-chip and c. elegans -based innovations.

1. Introduction

2. overview of microfluidic systems utilized as drug screening systems and how they are fabricated, 3. drug screening applications of various cell/organ and model organism-based biochips, 3.1. cell/organ chips for drug screening, 3.2. c. elegans chips for drug screening, 4. conclusions and perspectives, author contributions, institutional review board statement, informed consent statement, data availability statement, acknowledgments, conflicts of interest.

• Irwin, S. Drug Screening and Evaluative Procedures. Science 1962 , 136 , 123–128. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Hughes, J.P.; Rees, S.S.; Kalindjian, S.B.; Philpott, K.L. Principles of Early Drug Discovery. Br. J. Pharmacol. 2011 , 162 , 1239. [ Google Scholar ] [ CrossRef ]
• Wu, G.; Doberstein, S.K. HTS Technologies in Biopharmaceutical Discovery. Drug Discov. Today 2006 , 11 , 718–724. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Iversen, P.W.; Beck, B.; Chen, Y.-F.; Dere, W.; Devanarayan, V.; Eastwood, B.J.; Farmen, M.W.; Iturria, S.J.; Montrose, C.; Moore, R.A.; et al. HTS Assay Validation. In Assay Guidance Manual ; Eli Lilly & Company and the National Center for Advancing Translational Sciences: Bethesda, MD, USA, 2012; pp. 1–20. [ Google Scholar ]
• Murphy, B.; Yin, H.; Maris, J.M.; Kolb, E.A.; Gorlick, R.; Reynolds, C.P.; Kang, M.H.; Keir, S.T.; Kurmasheva, R.T.; Dvorchik, I.; et al. Evaluation of Alternative in Vivo Drug Screening Methodology: A Single Mouse Analysis. Cancer Res. 2016 , 76 , 5798–5809. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Lee, H.C.; Lin, C.Y.; Tsai, H.J. Zebrafish, an In Vivo Platform to Screen Drugs and Proteins for Biomedical Use. Pharmaceuticals 2021 , 14 , 500. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Dreiman, G.H.S.; Bictash, M.; Fish, P.V.; Griffin, L.; Svensson, F. Changing the HTS Paradigm: AI-Driven Iterative Screening for Hit Finding. SLAS Discov. 2021 , 26 , 257. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Szymański, P.; Markowicz, M.; Mikiciuk-Olasik, E. Adaptation of High-Throughput Screening in Drug Discovery—Toxicological Screening Tests. Int. J. Mol. Sci. 2012 , 13 , 427. [ Google Scholar ] [ CrossRef ]
• Butkiewicz, M.; Wang, Y.; Bryant, S.H.; Lowe, E.W., Jr.; Weaver, D.C.; Meiler, J. High-Throughput Screening Assay Datasets from the PubChem Database. Chem. Inform. 2017 , 3 , 1. [ Google Scholar ] [ CrossRef ]
• Szabo, M.; Akusjärvi, S.S.; Saxena, A.; Liu, J.; Chandrasekar, G.; Kitambi, S.S. Cell and Small Animal Models for Phenotypic Drug Discovery. Drug Des. Dev. Ther. 2017 , 11 , 1957–1967. [ Google Scholar ] [ CrossRef ]
• Fursov, N.; Cong, M.; Federici, M.; Platchek, M.; Haytko, P.; Tacke, R.; Livelli, T.; Zhong, Z. Improving Consistency of Cell-Based Assays by Using Division-Arrested Cells. Assay Drug Dev. Technol. 2005 , 3 , 7–15. [ Google Scholar ] [ CrossRef ]
• Tannenbaum, J.; Bennett, B.T. Russell and Burch’s 3Rs Then and Now: The Need for Clarity in Definition and Purpose. J. Am. Assoc. Lab. Anim. Sci. 2015 , 54 , 120. [ Google Scholar ] [ PubMed ]
• Van Norman, G.A. Limitations of Animal Studies for Predicting Toxicity in Clinical Trials: Is It Time to Rethink Our Current Approach? JACC Basic Transl. Sci. 2019 , 4 , 845. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Beebe, D.J.; Mensing, G.A.; Walker, G.M. Physics and Applications of Microfluidics in Biology. Annu. Rev. Biomed. Eng. 2002 , 4 , 261–286. [ Google Scholar ] [ CrossRef ]
• Masuda, S.; Washizu, M.; Nanba, T. Novel Method of Cell Fusion In Field Constriction Area In Fluid Integrated Circuit. IEEE Trans. Ind. Appl. 1989 , 25 , 732–737. [ Google Scholar ] [ CrossRef ]
• Du, G.; Fang, Q.; den Toonder, J.M.J. Microfluidics for Cell-Based High Throughput Screening Platforms—A Review. Anal. Chim. Acta 2016 , 903 , 36–50. [ Google Scholar ] [ CrossRef ]
• Chen, Y.; Chan, H.N.; Michael, S.A.; Shen, Y.; Chen, Y.; Tian, Q.; Huang, L.; Wu, H. A Microfluidic Circulatory System Integrated with Capillary-Assisted Pressure Sensors. Lab Chip 2017 , 17 , 653–662. [ Google Scholar ] [ CrossRef ]
• Tan, K.; Keegan, P.; Rogers, M.; Lu, M.; Gosset, J.R.; Charest, J.; Bale, S.S. A High-Throughput Microfluidic Microphysiological System (PREDICT-96) to Recapitulate Hepatocyte Function in Dynamic, Re-Circulating Flow Conditions. Lab Chip 2019 , 19 , 1556–1566. [ Google Scholar ] [ CrossRef ]
• Chramiec, A.; Teles, D.; Yeager, K.; Marturano-Kruik, A.; Pak, J.; Chen, T.; Hao, L.; Wang, M.; Lock, R.; Tavakol, D.N.; et al. Integrated Human Organ-on-a-Chip Model for Predictive Studies of Anti-Tumor Drug Efficacy and Cardiac Safety. Lab Chip 2020 , 20 , 4357–4372. [ Google Scholar ] [ CrossRef ]
• Kaletta, T.; Hengartner, M.O. Finding Function in Novel Targets: C. elegans as a Model Organism. Nat. Rev. Drug Discov. 2006 , 5 , 387–398. [ Google Scholar ] [ CrossRef ]
• Brenner, S. The Genetics of Caenorhabditis Elegans. Genetics 1974 , 77 , 71–94. [ Google Scholar ] [ CrossRef ]
• Fire, A.; Xu, S.; Montgomery, M.K.; Kostas, S.A.; Driver, S.E.; Mello, C.C. Potent and Specific Genetic Interference by Double-Stranded RNA in Caenorhabditis Elegans. Nature 1998 , 391 , 806–811. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Sönnichsen, B.; Koski, L.B.; Walsh, A.; Marschall, P.; Neumann, B.; Brehm, M.; Alleaume, A.M.; Artelt, J.; Bettencourt, P.; Cassin, E.; et al. Full-Genome RNAi Profiling of Early Embryogenesis in Caenorhabditis Elegans. Nature 2005 , 434 , 462–469. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Girard, L.R.; Fiedler, T.J.; Harris, T.W.; Carvalho, F.; Antoshechkin, I.; Han, M.; Sternberg, P.W.; Stein, L.D.; Chalfie, M. WormBook: The Online Review of Caenorhabditis Elegans Biology. Nucleic Acids Res. 2007 , 35 , D472–D475. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• C. elegans Sequencing Consortium. Genome Sequence of the Nematode C. elegans : A Platform for Investigating Biology. Science 1998 , 282 , 2012–2018. [ Google Scholar ] [ CrossRef ]
• Lockery, S. Channeling the Worm: Microfluidic Devices for Nematode Neurobiology. Nat. Methods 2007 , 4 , 691–692. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Hulme, S.E.; Shevkoplyas, S.S.; Samuel, A. Microfluidics: Streamlining Discovery in Worm Biology. Nat. Methods 2008 , 5 , 589–590. [ Google Scholar ] [ CrossRef ]
• Hwang, H.; Lu, H. Microfluidic Tools for Developmental Studies of Small Model Organisms—Nematodes, Fruit Flies, and Zebrafish. Biotechnol. J. 2013 , 8 , 192–205. [ Google Scholar ] [ CrossRef ]
• Mondal, S.; Hegarty, E.; Sahn, J.J.; Scott, L.L.; Gökçe, S.K.; Martin, C.; Ghorashian, N.; Satarasinghe, P.N.; Iyer, S.; Sae-Lee, W.; et al. High-Content Microfluidic Screening Platform Used to Identify Σ2R/Tmem97 Binding Ligands That Reduce Age-Dependent Neurodegeneration in C. elegans SC-APP Model. ACS Chem. Neurosci. 2018 , 9 , 1014–1026. [ Google Scholar ] [ CrossRef ]
• Yang, A.; Lin, X.; Liu, Z.; Duan, X.; Yuan, Y.; Zhang, J.; Liang, Q.; Ji, X.; Sun, N.; Yu, H.; et al. Worm Generator: A System for High-Throughput in Vivo Screening. Nano Lett. 2023 , 23 , 1280–1288. [ Google Scholar ] [ CrossRef ]
• Dong, L.; Jankele, R.; Cornaglia, M.; Lehnert, T.; Gönczy, P.; Gijs, M.A.M. Integrated Microfluidic Device for Drug Studies of Early C. elegans Embryogenesis. Adv. Sci. 2018 , 5 , 1700751. [ Google Scholar ] [ CrossRef ]
• Sofela, S.; Sahloul, S.; Song, Y.A. Biophysical Analysis of Drug Efficacy on C. Elegans Models for Neurodegenerative and Neuromuscular Diseases. PLoS ONE 2021 , 16 , e0246496. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Weeks, J.C.; Robinson, K.J.; Lockery, S.R.; Roberts, W.M. Anthelmintic Drug Actions in Resistant and Susceptible C. elegans Revealed by Electrophysiological Recordings in a Multichannel Microfluidic Device. Int. J. Parasitol. Drugs Drug Resist. 2018 , 8 , 607–628. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Xia, Y.; Whitesides, G.M. Soft Lithography. Annu. Rev. Mater. Sci. 1998 , 28 , 153–184. [ Google Scholar ] [ CrossRef ]
• Whitesides, G.M. The Origins and the Future of Microfluidics. Nature 2006 , 442 , 368–373. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Ma, C.; Peng, Y.; Li, H.; Chen, W. Organ-on-a-Chip: A New Paradigm for Drug Development. Trends Pharmacol. Sci. 2021 , 42 , 119. [ Google Scholar ] [ CrossRef ]
• Carr, J.A.; Parashar, A.; Gibson, R.; Robertson, A.P.; Martin, R.J.; Pandey, S. A Microfluidic Platform for High-Sensitivity, Real-Time Drug Screening on C. elegans and Parasitic Nematodes. Lab Chip 2011 , 11 , 2385–2396. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Prince, E.; Kheiri, S.; Wang, Y.; Xu, F.; Cruickshank, J.; Topolskaia, V.; Tao, H.; Young, E.W.K.; McGuigan, A.P.; Cescon, D.W.; et al. Microfluidic Arrays of Breast Tumor Spheroids for Drug Screening and Personalized Cancer Therapies. Adv. Healthc. Mater. 2022 , 11 , 2101085. [ Google Scholar ] [ CrossRef ]
• Sun, Q.; Tan, S.H.; Chen, Q.; Ran, R.; Hui, Y.; Chen, D.; Zhao, C.X. Microfluidic Formation of Coculture Tumor Spheroids with Stromal Cells As a Novel 3D Tumor Model for Drug Testing. ACS Biomater. Sci. Eng. 2018 , 4 , 4425–4433. [ Google Scholar ] [ CrossRef ]
• Yang, J.; Zheng, M.; Yang, F.; Zhang, X.; Yin, W.; Liu, X.; Zhang, G.J.; Chen, Z. Pseudomonas Aeruginosa Infected Nematode-on-a-Chip Model Array for Antibacterials Screening. Sens. Actuators B Chem. 2018 , 275 , 373–381. [ Google Scholar ] [ CrossRef ]
• Li, L.; Chen, Y.; Wang, H.; An, G.; Wu, H.; Huang, W. A High-Throughput, Open-Space and Reusable Microfluidic Chip for Combinational Drug Screening on Tumor Spheroids. Lab Chip 2021 , 21 , 3924–3932. [ Google Scholar ] [ CrossRef ]
• Tobias, F.; McIntosh, J.C.; Labonia, G.J.; Boyce, M.W.; Lockett, M.R.; Hummon, A.B. Developing a Drug Screening Platform: MALDI-Mass Spectrometry Imaging of Paper-Based Cultures. Anal. Chem. 2019 , 91 , 15370–15376. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Kim, S.; Ko, J.; Lee, S.R.; Park, D.; Park, S.; Jeon, N.L. Anchor-IMPACT: A Standardized Microfluidic Platform for High-Throughput Antiangiogenic Drug Screening. Biotechnol. Bioeng. 2021 , 118 , 2524–2535. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Choi, H.S.; Ahn, G.N.; Na, G.S.; Cha, H.J.; Kim, D.P. A Perfluoropolyether Microfluidic Device for Cell-Based Drug Screening with Accurate Quantitative Analysis. ACS Biomater. Sci. Eng. 2022 , 8 , 4577–4585. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Kulthong, K.; Duivenvoorde, L.; Sun, H.; Confederat, S.; Wu, J.; Spenkelink, B.; de Haan, L.; Marin, V.; van der Zande, M.; Bouwmeester, H. Microfluidic Chip for Culturing Intestinal Epithelial Cell Layers: Characterization and Comparison of Drug Transport between Dynamic and Static Models. Toxicol. In Vitr. 2020 , 65 , 104815. [ Google Scholar ] [ CrossRef ]
• Bhusal, A.; Dogan, E.; Nieto, D.; Mousavi Shaegh, S.A.; Cecen, B.; Miri, A.K. 3D Bioprinted Hydrogel Microfluidic Devices for Parallel Drug Screening. ACS Appl. Bio. Mater. 2022 , 2022 , 4492. [ Google Scholar ] [ CrossRef ]
• Markov, D.A.; Lillie, E.M.; Garbett, S.P.; McCawley, L.J. Variation in Diffusion of Gases through PDMS Due to Plasma Surface Treatment and Storage Conditions. Biomed. Microdevices 2014 , 16 , 91. [ Google Scholar ] [ CrossRef ]
• Whitesides, G.M.; Ostuni, E.; Takayama, S.; Jiang, X.; Ingber, D.E. Soft Lithography in Biology and Biochemistry. Annu. Rev. Biomed. Eng. 2003 , 3 , 335–373. [ Google Scholar ] [ CrossRef ]
• Abgrall, P.; Conedera, V.; Camon, H.; Gue, A.M.; Nguyen, N.T. SU-8 as a Structural Material for Labs-on-Chips and Microelectromechanical Systems. Electrophoresis 2007 , 28 , 4539–4551. [ Google Scholar ] [ CrossRef ]
• Wu, J.; Gao, Y.; Xi, J.; You, X.; Zhang, X.; Zhang, X.; Cao, Y.; Liu, P.; Chen, X.; Luan, Y. A High-Throughput Microplate Toxicity Screening Platform Based on Caenorhabditis Elegans. Ecotoxicol. Environ. Saf. 2022 , 245 , 114089. [ Google Scholar ] [ CrossRef ]
• Shirure, V.S.; George, S.C. Design Considerations to Minimize the Impact of Drug Absorption in Polymer-Based Organ-on-a-Chip Platforms. Lab Chip 2017 , 17 , 681–690. [ Google Scholar ] [ CrossRef ]
• van Meer, B.J.; de Vries, H.; Firth, K.S.A.; van Weerd, J.; Tertoolen, L.G.J.; Karperien, H.B.J.; Jonkheijm, P.; Denning, C.; IJzerman, A.P.; Mummery, C.L. Small Molecule Absorption by PDMS in the Context of Drug Response Bioassays. Biochem. Biophys. Res. Commun. 2017 , 482 , 323–328. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Sofela, S.; Sahloul, S.; Stubbs, C.; Orozaliev, A.; Refai, F.S.; Esmaeel, A.M.; Fahs, H.; Abdelgawad, M.O.; Gunsalus, K.C.; Song, Y.A. Phenotyping of the Thrashing Forces Exerted by Partially Immobilized C. elegans Using Elastomeric Micropillar Arrays. Lab Chip 2019 , 19 , 3685–3696. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Zhang, B.; Zhuang, L.; Sun, D.; Li, Y.; Chen, Z. An Integrated Microfluidics for Assessing the Anti-Aging Effect of Caffeic Acid Phenethylester in Caenorhabditis Elegans. Electrophoresis 2021 , 42 , 742–748. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Wang, C.; Zhao, S.; Zhao, X.; Chen, L.; Tian, Z.; Chen, X.; Qin, S. A Novel Wide-Range Microfluidic Dilution Device for Drug Screening. Biomicrofluidics 2019 , 13 , 024105. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Jahangiri, F.; Hakala, T.; Jokinen, V. Long-Term Hydrophilization of Polydimethylsiloxane (PDMS) for Capillary Filling Microfluidic Chips. Microfluid. Nanofluid. 2020 , 24 , 2. [ Google Scholar ] [ CrossRef ]
• Matellan, C.; Del Río Hernández, A.E. Cost-Effective Rapid Prototyping and Assembly of Poly(Methyl Methacrylate) Microfluidic Devices. Sci. Rep. 2018 , 8 , 6971. [ Google Scholar ] [ CrossRef ]
• Du, L.; Chang, H.; Song, M.; Liu, C. The Effect of Injection Molding PMMA Microfluidic Chips Thickness Uniformity on the Thermal Bonding Ratio of Chips. Microsyst. Technol. 2012 , 18 , 815–822. [ Google Scholar ] [ CrossRef ]
• Tweedie, M.; Maguire, P.D. Microfluidic Ratio Metering Devices Fabricated in PMMA by CO 2 Laser. Microsyst. Technol. 2021 , 27 , 47–58. [ Google Scholar ] [ CrossRef ]
• Mecomber, J.S.; Stalcup, A.M.; Hurd, D.; Halsall, H.B.; Heineman, W.R.; Seliskar, C.J.; Wehmeyer, K.R.; Limbach, P.A. Analytical Performance of Polymer-Based Microfluidic Devices Fabricated by Computer Numerical Controlled Machining. Anal. Chem. 2006 , 78 , 936–941. [ Google Scholar ] [ CrossRef ]
• Balaji, V.; Castro, K.; Folch, A. A Laser-Engraving Technique for Portable Micropneumatic Oscillators. Micromachines 2018 , 9 , 426. [ Google Scholar ] [ CrossRef ]
• Riester, O.; Laufer, S.; Deigner, H.P. Direct 3D Printed Biocompatible Microfluidics: Assessment of Human Mesenchymal Stem Cell Differentiation and Cytotoxic Drug Screening in a Dynamic Culture System. J. Nanobiotechnol. 2022 , 20 , 540. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Schuster, B.; Junkin, M.; Kashaf, S.S.; Romero-Calvo, I.; Kirby, K.; Matthews, J.; Weber, C.R.; Rzhetsky, A.; White, K.P.; Tay, S. Automated Microfluidic Platform for Dynamic and Combinatorial Drug Screening of Tumor Organoids. Nat. Commun. 2020 , 11 , 5271. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Zhao, S.K.; Hu, X.J.; Zhu, J.M.; Luo, Z.Y.; Liang, L.; Yang, D.Y.; Chen, Y.L.; Chen, L.F.; Zheng, Y.J.; Hu, Q.H.; et al. On-Chip Rapid Drug Screening of Leukemia Cells by Acoustic Streaming. Lab Chip 2021 , 21 , 4005–4015. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Eilenberger, C.; Rothbauer, M.; Selinger, F.; Gerhartl, A.; Jordan, C.; Harasek, M.; Schädl, B.; Grillari, J.; Weghuber, J.; Neuhaus, W.; et al. A Microfluidic Multisize Spheroid Array for Multiparametric Screening of Anticancer Drugs and Blood–Brain Barrier Transport Properties. Adv. Sci. 2021 , 8 , 2004856. [ Google Scholar ] [ CrossRef ]
• Chen, X.; Chen, H.; Wu, D.; Chen, Q.; Zhou, Z.; Zhang, R.; Peng, X.; Su, Y.C.; Sun, D. 3D Printed Microfluidic Chip for Multiple Anticancer Drug Combinations. Sens. Actuators B Chem. 2018 , 276 , 507–516. [ Google Scholar ] [ CrossRef ]
• Jaberi, A.; Monemian Esfahani, A.; Aghabaglou, F.; Park, J.S.; Ndao, S.; Tamayol, A.; Yang, R. Microfluidic Systems with Embedded Cell Culture Chambers for High-Throughput Biological Assays. ACS Appl. Bio. Mater. 2020 , 3 , 6661–6671. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Mi, S.; Liu, Z.; Du, Z.; Yi, X.; Sun, W. Three-Dimensional Microfluidic Tumor–Macrophage System for Breast Cancer Cell Invasion. Biotechnol. Bioeng. 2019 , 116 , 1731–1741. [ Google Scholar ] [ CrossRef ]
• Haque, M.R.; Wessel, C.R.; Leary, D.D.; Wang, C.; Bhushan, A.; Bishehsari, F. Patient-Derived Pancreatic Cancer-on-a-Chip Recapitulates the Tumor Microenvironment. Microsyst. Nanoeng. 2022 , 8 , 36. [ Google Scholar ] [ CrossRef ]
• Yin, L.; Du, G.; Zhang, B.; Zhang, H.; Yin, R.; Zhang, W.; Yang, S.M. Efficient Drug Screening and Nephrotoxicity Assessment on Co-Culture Microfluidic Kidney Chip. Sci. Rep. 2020 , 10 , 6568. [ Google Scholar ] [ CrossRef ]
• Fantuzzo, J.A.; Robles, D.A.; Mirabella, V.R.; Hart, R.P.; Pang, Z.P.; Zahn, J.D. Development of a High-Throughput Arrayed Neural Circuitry Platform Using Human Induced Neurons for Drug Screening Applications. Lab Chip 2020 , 20 , 1140–1152. [ Google Scholar ] [ CrossRef ]
• Chen, Y.; Gao, D.; Wang, Y.; Lin, S.; Jiang, Y. A Novel 3D Breast-Cancer-on-Chip Platform for Therapeutic Evaluation of Drug Delivery Systems. Anal. Chim. Acta 2018 , 1036 , 97–106. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Mitxelena-Iribarren, O.; Zabalo, J.; Arana, S.; Mujika, M. Improved Microfluidic Platform for Simultaneous Multiple Drug Screening towards Personalized Treatment. Biosens. Bioelectron. 2019 , 123 , 237–243. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Zhang, J.; Chintalaramulu, N.; Vadivelu, R.; An, H.; Yuan, D.; Jin, J.; Ooi, C.H.; Cock, I.E.; Li, W.; Nguyen, N.T. Inertial Microfluidic Purification of Floating Cancer Cells for Drug Screening and Three-Dimensional Tumor Models. Anal. Chem. 2020 , 92 , 11558–11564. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Lin, D.; Li, P.; Feng, J.; Lin, Z.; Chen, X.; Yang, N.; Wang, L.; Liu, D. Screening Therapeutic Agents Specific to Breast Cancer Stem Cells Using a Microfluidic Single-Cell Clone-Forming Inhibition Assay. Small 2020 , 16 , 1901001. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Zhang, H.; Xiao, L.; Li, Q.; Qi, X.; Zhou, A. Microfluidic Chip for Non-Invasive Analysis of Tumor Cells Interaction with Anti-Cancer Drug Doxorubicin by AFM and Raman Spectroscopy. Biomicrofluidics 2018 , 12 , 55. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Moon, H.S.; Yoo, C.E.; Kim, S.; Lee, J.E.; Park, W.Y. Application of an Open-Chamber Multi-Channel Microfluidic Device to Test Chemotherapy Drugs. Sci. Rep. 2020 , 10 , 20343. [ Google Scholar ] [ CrossRef ]
• Chang, S.; Wen, J.; Su, Y.; Ma, H. Microfluidic Platform for Studying the Anti-Cancer Effect of Ursolic Acid on Tumor Spheroid. Electrophoresis 2022 , 43 , 1466–1475. [ Google Scholar ] [ CrossRef ]
• Jang, S.D.; Song, J.; Kim, H.A.; Im, C.N.; Khawar, I.A.; Park, J.K.; Kuh, H.J. Anti-Cancer Activity Profiling of Chemotherapeutic Agents in 3D Co-Cultures of Pancreatic Tumor Spheroids with Cancer- Associated Fibroblasts and Macrophages. Cancers 2021 , 13 , 5955. [ Google Scholar ] [ CrossRef ]
• Sun, Y.J.; Hsu, C.H.; Ling, T.Y.; Liu, L.; Lin, T.C.; Jakfar, S.; Young, I.C.; Lin, F.H. The Preparation of Cell-Containing Microbubble Scaffolds to Mimic Alveoli Structure as a 3D Drug-Screening System for Lung Cancer. Biofabrication 2020 , 12 , 025031. [ Google Scholar ] [ CrossRef ]
• Sankar, S.; Mehta, V.; Ravi, S.; Sharma, C.S.; Rath, S.N. A Novel Design of Microfluidic Platform for Metronomic Combinatorial Chemotherapy Drug Screening Based on 3D Tumor Spheroid Model. Biomed. Microdevices 2021 , 23 , 50. [ Google Scholar ] [ CrossRef ]
• Khot, M.I.; Levenstein, M.A.; de Boer, G.N.; Armstrong, G.; Maisey, T.; Svavarsdottir, H.S.; Andrew, H.; Perry, S.L.; Kapur, N.; Jayne, D.G. Characterising a PDMS Based 3D Cell Culturing Microfluidic Platform for Screening Chemotherapeutic Drug Cytotoxic Activity. Sci. Rep. 2020 , 10 , 15915. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Ma, H.P.; Deng, X.; Chen, D.Y.; Zhu, D.; Tong, J.L.; Zhao, T.; Ma, J.H.; Liu, Y.Q. A Microfluidic Chip-Based Co-Culture of Fibroblast-like Synoviocytes with Osteoblasts and Osteoclasts to Test Bone Erosion and Drug Evaluation. R. Soc. Open Sci. 2018 , 5 , 180528. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Shi, Y.; He, X.; Wang, H.; Dai, J.; Fang, J.; He, Y.; Chen, X.; Hong, Z.; Chai, Y. Construction of a Novel Blood Brain Barrier-Glioma Microfluidic Chip Model: Applications in the Evaluation of Permeability and Anti-Glioma Activity of Traditional Chinese Medicine Components. Talanta 2023 , 253 , 123971. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Clancy, A.; Chen, D.; Bruns, J.; Nadella, J.; Stealey, S.; Zhang, Y.; Timperman, A.; Zustiak, S.P. Hydrogel-Based Microfluidic Device with Multiplexed 3D in Vitro Cell Culture. Sci. Rep. 2022 , 12 , 17781. [ Google Scholar ] [ CrossRef ]
• Ming, L.; Zhipeng, Y.; Fei, Y.; Feng, R.; Jian, W.; Baoguo, J.; Yongqiang, W.; Peixun, Z. Microfluidic-Based Screening of Resveratrol and Drug-Loading PLA/Gelatine Nano-Scaffold for the Repair of Cartilage Defect. Artif. Cells Nanomed. Biotechnol. 2018 , 46 , 336–346. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Mulholland, T.; McAllister, M.; Patek, S.; Flint, D.; Underwood, M.; Sim, A.; Edwards, J.; Zagnoni, M. Drug Screening of Biopsy-Derived Spheroids Using a Self-Generated Microfluidic Concentration Gradient. Sci. Rep. 2018 , 8 , 14672. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Jin, Y.; Kim, J.; Lee, J.S.; Min, S.; Kim, S.; Ahn, D.H.; Kim, Y.G.; Cho, S.W. Vascularized Liver Organoids Generated Using Induced Hepatic Tissue and Dynamic Liver-Specific Microenvironment as a Drug Testing Platform. Adv. Funct. Mater. 2018 , 28 , 1801954. [ Google Scholar ] [ CrossRef ]
• Mo, S.J.; Lee, J.H.; Kye, H.G.; Lee, J.M.; Kim, E.J.; Geum, D.; Sun, W.; Chung, B.G. A Microfluidic Gradient Device for Drug Screening with Human IPSC-Derived Motoneurons. Analyst 2020 , 145 , 3081–3089. [ Google Scholar ] [ CrossRef ]
• Tang, Q.; Li, X.; Lai, C.; Li, L.; Wu, H.; Wang, Y.; Shi, X. Fabrication of a Hydroxyapatite-PDMS Microfluidic Chip for Bone-Related Cell Culture and Drug Screening. Bioact. Mater. 2021 , 6 , 169–178. [ Google Scholar ] [ CrossRef ]
• Ko, J.; Ahn, J.; Kim, S.; Lee, Y.; Lee, J.; Park, D.; Jeon, N.L. Tumor Spheroid-on-a-Chip: A Standardized Microfluidic Culture Platform for Investigating Tumor Angiogenesis. Lab Chip 2019 , 19 , 2822–2833. [ Google Scholar ] [ CrossRef ]
• Zhai, J.; Li, C.; Li, H.; Yi, S.; Yang, N.; Miao, K.; Deng, C.; Jia, Y.; Mak, P.I.; Martins, R.P. Cancer Drug Screening with an On-Chip Multi-Drug Dispenser in Digital Microfluidics. Lab Chip 2021 , 21 , 4749–4759. [ Google Scholar ] [ CrossRef ]
• Akay, M.; Hite, J.; Avci, N.G.; Fan, Y.; Akay, Y.; Lu, G.; Zhu, J.J. Drug Screening of Human GBM Spheroids in Brain Cancer Chip. Sci. Rep. 2018 , 8 , 15423. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Atakan, H.B.; Xiang, R.; Cornaglia, M.; Mouchiroud, L.; Katsyuba, E.; Auwerx, J.; Gijs, M.A.M. Automated Platform for Long-Term Culture and High-Content Phenotyping of Single C. elegans Worms. Sci. Rep. 2019 , 9 , 14340. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Sun, G.; Manning, C.A.; Lee, G.H.; Majeed, M.; Lu, H. Microswimmer Combing: Controlling Interfacial Dynamics for Open-Surface Multifunctional Screening of Small Animals. Adv. Healthc. Mater. 2021 , 10 , 2001887. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Letizia, M.C.; Cornaglia, M.; Tranchida, G.; Trouillon, R.; Gijs, M.A.M. A Design of Experiment Approach for Efficient Multi-Parametric Drug Testing Using a Caenorhabditis Elegans Model. Integr. Biol. 2018 , 10 , 48–56. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Banse, S.A.; Blue, B.W.; Robinson, K.J.; Jarrett, C.M.; Phillips, P.C. The Stress-Chip: A Microfluidic Platform for Stress Analysis in Caenorhabditis Elegans. PLoS ONE 2019 , 14 , e0216283. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Hu, L.; Ge, A.; Wang, X.; Wang, S.; Yue, X.; Wang, J.; Feng, X.; Du, W.; Liu, B.F. Real-Time Monitoring of Immune Responses under Pathogen Invasion and Drug Interference by Integrated Microfluidic Device Coupled with Worm-Based Biosensor. Biosens. Bioelectron. 2018 , 110 , 233–238. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Atakan, H.B.; Hof, K.S.; Cornaglia, M.; Auwerx, J.; Gijs, M.A.M. The Detection of Early Epigenetic Inheritance of Mitochondrial Stress in C. elegans with a Microfluidic Phenotyping Platform. Sci. Rep. 2019 , 9 , 19315. [ Google Scholar ] [ CrossRef ]
• Rezai, P.; Siddiqui, A.; Selvaganapathy, P.R.; Gupta, B.P. Electrotaxis of Caenorhabditis Elegans in a Microfluidic Environment. Lab Chip 2010 , 10 , 220–226. [ Google Scholar ] [ CrossRef ]
• Dwyer, N.; Adler, E.C.; Crump, G.J.; L’Etoile, D.N.; Bargmann, I.C. Polarized Dendritic Transport and the AP-1 Μ1 Clathrin Adaptor UNC-101 Localize Odorant Receptors to Olfactory Cilia. Neuron 2001 , 31 , 277–287. [ Google Scholar ] [ CrossRef ]
• Fang-Yen, C.; Gabel, C.V.; Samuel, A.D.; Bargmann, C.I.; Avery, L. Laser Microsurgery in Caenorhabditis Elegans. Methods Cell Biol. 2012 , 107 , 177–206. [ Google Scholar ] [ PubMed ]
• Kerr, R.; Lev-Ram, V.; Baird, G.; Vincent, P.; Tsien, R.Y.; Schafer, W.R. Optical Imaging of Calcium Transients in Neurons and Pharyngeal Muscle of C. elegans . Neuron 2000 , 26 , 583–594. [ Google Scholar ] [ CrossRef ]
• Goodman, M.B.; Hall, D.H.; Avery, L.; Lockery, S.R. Active Currents Regulate Sensitivity and Dynamic Range in C. elegans Neurons. Neuron 1998 , 20 , 763–772. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Mondal, S.; Ahlawat, S.; Koushika, S.P. Simple Microfluidic Devices for in Vivo Imaging of C. elegans , Drosophila and Zebrafish. J. Vis. Exp. 2012 , 67 , e3780. [ Google Scholar ]
• Cooksey, G.A.; Atencia, J. Pneumatic Valves in Folded 2D and 3D Fluidic Devices Made from Plastic Films and Tapes. Lab Chip 2014 , 14 , 1665–1668. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Krajniak, J.; Lu, H. Long-Term High-Resolution Imaging and Culture of C. elegans in Chip-Gel Hybrid Microfludic Device for Developmental Studies. Lab Chip 2010 , 10 , 1862. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Cho, Y.; Ah Lee, S.; Lian Chew, Y.; Broderick, K.; Schafer, W.R.; Lu, H.; Cho, Y.; Lee, S.A.; Broderick, K.; Lu, H.; et al. Multimodal Stimulation in a Microfluidic Device Facilitates Studies of Interneurons in Sensory Integration in C. elegans . Small 2020 , 16 , 1905852. [ Google Scholar ] [ CrossRef ]
• Migliozzi, D.; Cornaglia, M.; Mouchiroud, L.; Uhlmann, V.; Unser, M.A.; Auwerx, J.; Gijs, M.A.M. Multimodal Imaging and High-Throughput Image-Processing for Drug Screening on Living Organisms on-Chip. J. Biomed. Opt. 2019 , 24 , 021205. [ Google Scholar ] [ CrossRef ]
• Chen, Z.; Deng, J.; Zhang, X.; Luo, Y.; Lu, Y.; Wu, Z.; Lin, B. A Novel Micro-Injection Droplet Microfluidic System for Studying Locomotive Behavior Responses to Cu 2+ Induced Neurotoxin in Individual C. elegans . Anal. Chim. Acta 2020 , 1106 , 61–70. [ Google Scholar ] [ CrossRef ]
• Letizia, M.C.; Cornaglia, M.; Trouillon, R.; Sorrentino, V.; Mouchiroud, L.; Bou Sleiman, M.S.; Auwerx, J.; Gijs, M.A.M. Microfluidics-Enabled Phenotyping of a Whole Population of C. elegans Worms over Their Embryonic and Post-Embryonic Development at Single-Organism Resolution. Microsyst. Nanoeng. 2018 , 4 , 6. [ Google Scholar ] [ CrossRef ]
• Paul, D.; Sanap, G.; Shenoy, S.; Kalyane, D.; Kalia, K.; Tekade, R.K. Artificial Intelligence in Drug Discovery and Development. Drug Discov. Today 2021 , 26 , 80. [ Google Scholar ] [ CrossRef ] [ PubMed ]

Click here to enlarge figure

Share and Cite

Yoon, S.; Kilicarslan You, D.; Jeong, U.; Lee, M.; Kim, E.; Jeon, T.-J.; Kim, S.M. Microfluidics in High-Throughput Drug Screening: Organ-on-a-Chip and C. elegans -Based Innovations. Biosensors 2024 , 14 , 55. https://doi.org/10.3390/bios14010055

Yoon S, Kilicarslan You D, Jeong U, Lee M, Kim E, Jeon T-J, Kim SM. Microfluidics in High-Throughput Drug Screening: Organ-on-a-Chip and C. elegans -Based Innovations. Biosensors . 2024; 14(1):55. https://doi.org/10.3390/bios14010055

Yoon, Sunhee, Dilara Kilicarslan You, Uiechan Jeong, Mina Lee, Eunhye Kim, Tae-Joon Jeon, and Sun Min Kim. 2024. "Microfluidics in High-Throughput Drug Screening: Organ-on-a-Chip and C. elegans -Based Innovations" Biosensors 14, no. 1: 55. https://doi.org/10.3390/bios14010055

Article Metrics

Article access statistics, further information, mdpi initiatives, follow mdpi.

Subscribe to receive issue release notifications and newsletters from MDPI journals

Drug Screening - Science topic

• Recruit researchers
• Join for free
• Login Email Tip: Most researchers use their institutional email address as their ResearchGate login Password Forgot password? Keep me logged in Log in or Continue with Google Welcome back! Please log in. Email · Hint Tip: Most researchers use their institutional email address as their ResearchGate login Password Forgot password? Keep me logged in Log in or Continue with Google No account? Sign up

MINI REVIEW article

Qsar-based virtual screening: advances and applications in drug discovery.

• 1 LabMol – Laboratory for Molecular Modeling and Drug Design, Faculdade de Farmácia, Universidade Federal de Goiás, Goiânia, Brazil
• 2 Laboratory of Cheminformatics, Centro Universitário de Anápolis (UniEVANGÉLICA), Anápolis, Brazil
• 3 Laboratory for Molecular Modeling, Division of Chemical Biology and Medicinal Chemistry, Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States
• 4 Department of Chemical Technology, Odessa National Polytechnic University, Odessa, Ukraine

Virtual screening (VS) has emerged in drug discovery as a powerful computational approach to screen large libraries of small molecules for new hits with desired properties that can then be tested experimentally. Similar to other computational approaches, VS intention is not to replace in vitro or in vivo assays, but to speed up the discovery process, to reduce the number of candidates to be tested experimentally, and to rationalize their choice. Moreover, VS has become very popular in pharmaceutical companies and academic organizations due to its time-, cost-, resources-, and labor-saving. Among the VS approaches, quantitative structure–activity relationship (QSAR) analysis is the most powerful method due to its high and fast throughput and good hit rate. As the first preliminary step of a QSAR model development, relevant chemogenomics data are collected from databases and the literature. Then, chemical descriptors are calculated on different levels of representation of molecular structure, ranging from 1D to n D, and then correlated with the biological property using machine learning techniques. Once developed and validated, QSAR models are applied to predict the biological property of novel compounds. Although the experimental testing of computational hits is not an inherent part of QSAR methodology, it is highly desired and should be performed as an ultimate validation of developed models. In this mini-review, we summarize and critically analyze the recent trends of QSAR-based VS in drug discovery and demonstrate successful applications in identifying perspective compounds with desired properties. Moreover, we provide some recommendations about the best practices for QSAR-based VS along with the future perspectives of this approach.

Introduction

Quantitative structure–activity relationship (QSAR) analysis is a ligand-based drug design method developed more than 50 years ago by Hansch and Fujita (1964) . Since then and until now, QSAR remains an efficient method for building mathematical models, which attempts to find a statistically significant correlation between the chemical structure and continuous (pIC 50 , pEC 50 , Ki, etc.) or categorical/binary (active, inactive, toxic, nontoxic, etc.) biological/toxicological property using regression and classification techniques, respectively ( Cherkasov et al., 2014 ). In the last decades, QSAR has undergone several transformations, ranging from the dimensionality of the molecular descriptors (from 1D to n D) and different methods for finding a correlation between the chemical structures and the biological property. Initially, QSAR modeling was limited to small series of congeneric compounds and simple regression methods. Nowadays, QSAR modeling has grown, diversified, and evolved to the modeling and virtual screening (VS) of very large data sets comprising thousands of diverse chemical structures and using a wide variety of machine learning techniques ( Cherkasov et al., 2014 ; Mitchell, 2014 ; Ekins et al., 2015 ; Goh et al., 2017 ).

This review is devoted to (i) critical analysis of advantages and disadvantages of QSAR-based VS in drug discovery; (ii) demonstration of several successful QSAR-based discoveries of compounds with desired properties; (iii) description of best practices for the QSAR-based VS; and (iv) discussion of future perspectives of this approach.

Best Practices in QSAR Modeling and Validation

High-throughput screening (HTS) technologies resulted in the explosion of amount of data suitable for QSAR modeling. As a result, data quality problem became one of the fundamental questions in cheminformatics. As obvious as it seems, various errors in both chemical structure and experimental results are considered as major obstacle to building predictive models ( Young et al., 2008 ; Southan et al., 2009 ; Williams and Ekins, 2011 ).

Considering these limitations, Fourches et al. (2010 ; 2015 ; 2016 ) developed the guidelines for chemical and biological data curation as a first and mandatory step of the predictive QSAR modeling. Organized into a solid functional process, these guidelines allow the identification, correction, or, if needed, removal of structural and biological errors in large data sets. Data curation procedures include the removal of organometallics, counterions, mixtures, and inorganics, as well as the normalization of specific chemotypes, structural cleaning (e.g., detection of valence violations), standardization of tautomeric forms, and ring aromatization. Additional curation elements include averaging, aggregating, or removal of duplicates to produce a single bioactivity result. Detailed discussion of aforementioned data curation procedures can be found elsewhere ( Fourches et al., 2010 , 2015 , 2016 ).

The Organization for Economic Cooperation and Development (OECD) developed a set of guidelines that the researchers should follow to achieve the regulatory acceptance of QSAR models. According to these principles, QSAR models should be associated with (i) defined end point, (ii) unambiguous algorithm, (iii) defined domain of applicability, (iv) appropriate measures of goodness-of-fit, robustness, and predictivity, and (v) if possible, mechanistic interpretation ( OECD, 2004 ). In our opinion, the additional rule requesting thorough data curation as a mandatory preliminary step to model development should be added there.

Continuing Importance of QSAR as Virtual Screening Tool

The current pipeline to discover hit compounds in early stages of drug discovery is a data-driven process, which relies on bioactivity data obtained from HTS campaigns ( Nantasenamat and Prachayasittikul, 2015 ). Since the cost of obtaining new hit compounds in HTS platforms is rather high, QSAR modeling has been playing a pivotal role in prioritizing compounds for synthesis and/or biological evaluation. The QSAR models can be used for both hits identification and hit-to-lead optimization. In the latter, a favorable balance between potency, selectivity, and pharmacokinetic and toxicological parameters, which is required to develop a new, safe, and effective drug, could be achieved through several optimization cycles. As no compound need to be synthesized or tested before computational evaluation, QSAR represents a labor-, time-, and cost-effective method to obtain compounds with desired biological properties. Consequently, QSAR is widely practiced in industries, universities, and research centers around the world ( Cherkasov et al., 2014 ).

The general scheme of QSAR-based VS approach is shown in Figure 1 . Initially, the data sets collected from external sources are curated and integrated to remove or correct inconsistent data. Using these data, QSAR models are developed and validated following OECD guidelines and best practices of modeling. Then, QSAR models are used to identify chemical compounds predicted to be active against selected endpoints from large chemical libraries ( Cherkasov et al., 2014 ). In principle, VS is often compared to a funnel, where a large chemical library (i.e., 10 5 to 10 7 chemical structures) is reduced by QSAR models to a smaller number of compounds, which then will be tested experimentally (i.e., 10 1 to 10 3 chemical structures) ( Kar and Roy, 2013 ; Tanrikulu et al., 2013 ). However, it is important to mention that modern VS workflows incorporate additional filtering steps, including: (i) sets of empirical rules [e.g., Lipinski’s ( Lipinski et al., 1997 ) rules], (ii) chemical similarity cutoffs, (iii) other QSAR-based filters (e.g., toxicological and pharmacokinetic endpoints), and (iv) chemical feasibility and/or purchasability ( Cherkasov et al., 2014 ). Although the experimental validation of computational hits does not represent part of the QSAR methodology, this should be performed as the final important step. After experimental validation, a multi-parameter optimization (MPO) with QSAR predictions of potency, selectivity, and pharmacokinetic parameters can be conducted. This information will be crucial during hit-to lead and lead optimization design of the compound series, to find the properties balance (potency, selectivity, and PK) related with the effect of different decoration patterns to establish a new series of target compounds for in vivo evaluation.

FIGURE 1. QSAR-based virtual screening workflow.

QSAR-Based Virtual Screening vs. High-Throughput Screening

High-throughput screening can rapidly identify large subsets of molecules with desired activity from large screening collections of compounds (10 5 –10 6 compounds) using automated plate-based experimental assays ( Mueller et al., 2012 ). However, the hit rate of HTS ranges between 0.01% and 0.1% and this highlights the frequently encountered limitation that most of the screened compounds are routinely reported as inactive toward the desired bioactivity ( Thorne et al., 2010 ). Consequently, the drug discovery cost increases according to the number of tested compounds ( Butkiewicz et al., 2013 ). On the other hand, typical hit rates from a validated VS method, including QSAR-based, typically range between 1% and 40%. Thus, VS campaigns are found to have a higher rate of biologically active compounds and at a lower cost than HTS.

In this perspective, we show that QSAR-based VS could be used to enrich hit rates of HTS campaigns. For example, Mueller et al. (2010) employed both HTS and QSAR models to search novel positive allosteric modulators for mGlu 5 , a G-protein coupled receptor involved in disorders like schizophrenia and Parkinson’s disease. First, the HTS of approximately 144,000 compounds resulted in a total of 1,356 hits, with a hit rate of 0.94%. Then, this dataset was used to build continuous QSAR models (combining physicochemical descriptors and neural networks), which were subsequently applied to screen a database of approximately 450,000 compounds. Finally, 824 compounds were acquired for biological testing and 232 were confirmed as active (hit rate of 28.2%) ( Mueller et al., 2010 ). In another study, Rodriguez et al. (2010) screened approximately 160,000 compounds to identify 624 antagonists of mGlu 5 . Further, these data were used to develop QSAR models and, then, applied to screen near 700,000 compounds from ChemDiv database. Among them, 88 of acquired compounds were active, corresponding to a hit rate of 3.6% while the HTS had a hit rate of 0.2% ( Mueller et al., 2012 ).

Practical Applications of QSAR-Based Virtual Screening

Despite its obvious advantages, QSAR modeling remains underestimated as a VS tool. Unfortunately, QSAR is still seen as a complementary analysis to studies of synthesis and biological evaluation, often introduced in the study without any justification or additional perspective. Despite the small number of VS applications available in the literature, most of them led to the discovery of promising hits and lead candidates. Below, we discuss some successful applications of QSAR-based VS for the discovery of new hits and hit-to-lead optimization.

Malaria is an infectious disease caused by five different species of Plasmodium parasites and transmitted to humans through the bite of infected female mosquitoes of the genus Anopheles . The most lethal species is P. falciparum , which can lead to severe illness and death ( Phillips et al., 2017 ). Malaria is a widespread disease; 91 countries and areas have ongoing transmission. According to World Health Organization (WHO), about 216 million cases and 445,000 deaths from malaria were reported in 2016 ( WHO, 2018c ). Furthermore, the resistance to antimalarial drugs is a common and growing issue and constitutes a substantial threat for populations in endemic regions ( Gorobets et al., 2017 ; Menard and Dondorp, 2017 ). In a study reported by Zhang et al. (2013) , a data set of 3,133 compounds reported as active or inactive against P. falciparum chloroquine susceptible strain (3D7) was used to develop QSAR models. The models were built using Dragon descriptors (0D, 1D, and 2D), ISIDA-2D fragments descriptors and support vector machines (SVM) method. During QSAR modeling and validation, the data set was randomly divided into modeling and external evaluation set. Additionally, the modeling set was divided multiple times in training and test sets using the Sphere Exclusion algorithm. Then, by using a consensus approach, the QSAR models were applied for VS of the ChemBridge database. After VS, 176 potential antimalarial compounds were identified and submitted to experimental validation along with 42 putative inactive compounds, used as negative controls. Twenty-five compounds presented antimalarial activity in P. falciparum growth inhibition assays and low cytotoxicity in mammalian cells. All 42 compounds predicted as inactives by the models were confirmed experimentally ( Zhang et al., 2013 ). The confirmed experimental hits presented new chemical scaffolds against P. falciparum and could be promising starting points for the development of new optimized antimalarial agents.

Schistosomiasis

Schistosomiasis is a disease caused by flatworms of the genus Schistosoma that affects 206 million of people worldwide ( WHO, 2018d ). The current reliance on only one drug, praziquantel, for treatment and control of this disease calls for the urgent discovery of novel anti-schistosomal drugs ( Colley et al., 2014 ). Aiming at discovering new drugs, our group developed binary QSAR models for Schistosoma mansoni thioredoxin glutathione reductase ( Sm TGR), a validated target for schistosomiasis ( Kuntz et al., 2007 ), to find new structurally dissimilar compounds with antischistosomal activity ( Neves et al., 2016 ). To achieve this goal, we designed a study with the following steps: (i) curation of the largest possible data set of Sm TGR inhibitors, (ii) development of rigorously validated and mechanistically interpretable models, and (iii) application of generated models for VS of ChemBridge library. Using the QSAR models, we prioritized 29 compounds for further experimental evaluation. As a result, we found that the QSAR models were efficient for discovery of six novel hit compounds active against schistosomula and three hits active against adult worms (hit rate of 20.6%). Among them, 2-[2-(3-methyl-4-nitro-5-isoxazolyl)vinyl]pyridine and 2-(benzylsulfonyl)-1,3-benzothiazole, two compounds representing new chemical scaffolds have activity against schistosomula and adult worms at low micromolar concentrations and therefore represent promising antischistosomal hits for further hit-to-lead optimization ( Neves et al., 2016 ).

In another study, we developed continuous QSAR models for a data set of oxadiazoles inhibitors of sm TGR ( Melo-Filho et al., 2016 ). Using a combi-QSAR approach, we built a consensus model combining the predictions of individual 2D- and 3D-QSAR models. Then, the model was used for VS of ChemBridge database and the 10 top ranked compounds were further evaluated in vitro against schistosomula and adult worms. Additionally, we applied five highly predictive in-house QSAR models for prediction of important pharmacokinetics and toxicity properties of the new hits. The experimental results showed that 4-nitro-3,5-bis(1-nitro-1H-pyrazol-4-yl)-1H-pyrazole (LabMol-17) and 3-nitro-4-{[(4-nitro-1,2,5-oxadiazol-3-yl)oxy]methyl}-1,2,5-oxadiazole (LabMol-19), two compounds containing new chemical scaffolds (hit rate of 20.6%), were highly active in both life stages of the parasite at low micromolar concentrations ( Melo-Filho et al., 2016 ).

Tuberculosis

Mycobacterium tuberculosis , the causative agent of tuberculosis (TB), kills about 1.6 million people every year ( WHO, 2018e ). The current treatment of this disease takes approximately 9 months, which normally leads to noncompliance and, hence, the emergence of multidrug-resistant bacteria ( AlMatar et al., 2017 ). Aiming the design of new anti-TB agents, our group used QSAR models to design new series of chalcone (1,3-diaryl-2-propen-1-ones) derivatives. Initially, we retrieved from the literature all chalcone compounds with in vitro inhibition data against M. tuberculosis H37Rv strain. After rigorous data curation, these chalcones were subject to structure–activity relationships (SAR) analysis. Based on SAR rules, bioisosteric replacements were employed to design new chalcone derivatives with optimized anti-TB activity. In parallel, binary QSAR models were generated using several machine learning methods and molecular fingerprints. The fivefold external cross-validation procedure confirmed the high predictive power of the developed models. Using these models, we prioritized series of chalcone derivatives for synthesis and biological evaluation ( Gomes et al., 2017 ). As a result, five 5-nitro-substituted heteroaryl chalcones were found to exhibit MICs at nanomolar concentrations against replicating mycobacteria, as well as low micromolar activity against nonreplicating bacteria. In addition, four of these compounds were more potent than standard drug isoniazid. The series also showed low cytotoxicity against commensal bacteria and mammalian cells. These results suggest that designed heteroaryl chalcones, identified with the help of QSAR models, are promising anti-TB lead candidates ( Gomes et al., 2017 ).

Viral Infections

Yearly, influenza epidemics can seriously affect all populations in the world. These annual epidemics are estimated to result in about 5 million cases and 650,000 deaths ( WHO, 2018b ). Influenza virus is mutating constantly, resulting in novel resistant strains, and hence, the development of new anti-influenza drugs active against these new strains is important to prevent pandemics ( Laborda et al., 2016 ). Aiming the discovery of new anti-influenza A drugs, Lian et al. (2015) built binary QSAR models, using SVM and Naïve Bayesian methods, to predict neuraminidase inhibition, a validated protein target for influenza. Then, four different combinations of machine learning methods and molecular descriptors were applied to screen 15,600 compounds from an in-house database, among which 60 compounds were selected to experimental evaluation on neuraminidase activity. Nine inhibitors were identified, five of which were oseltamivir derivatives exhibiting potent neuraminidase inhibition at nanomolar concentrations. Other four active compounds belonged to novel scaffolds, with potent inhibition at low micromolar concentrations ( Lian et al., 2015 ).

According to WHO, approximately 35 million people are infected with HIV ( WHO, 2018a ). The treatment for HIV infections requires a lifelong antiretroviral therapy, targeting different stages of HIV replication cycle. Consequently, because of the emergence of resistance and the lack of tolerability, development of novel anti-HIV drugs is of high demand ( Cihlar and Fordyce, 2016 ; Garbelli et al., 2017 ). With the purpose of discovering new anti-HIV-1 drugs, Kurczyk et al. (2015) developed a two-step VS approach to prioritize compounds against HIV integrase, an important target to viral replication cycle. The first step was based on binary QSAR models, and the second on privileged fragments. Then, 1.5 million of commercially available compounds were screened, and 13 compounds were selected to be tested in vitro for inhibiting HIV-1 replication. Among them, two novel chemotypes with moderate anti-HIV-1 potencies were identified, and therefore, represent new starting points for prospective structural optimization studies.

Mood and Anxiety Disorders

The 5-hydroxytryptamine 1A (5-HT 1A ) serotonin receptor has been an attractive target for treating mood and anxiety disorders such as schizophrenia ( Nichols and Nichols, 2008 ; Lacivita et al., 2012 ). However, the currently marketed drugs targeting 5-HT 1A receptor possess severe side effects. To address this, Luo et al. (2014) developed a QSAR-based VS workflow to find new hit compounds targeting 5-HT 1A receptor. First, binary QSAR models were generated using Dragon descriptors and several machine learning methods. Then, developed QSAR models were rigorously validated and applied in consensus for VS four commercial chemical databases. Fifteen compounds were selected for experimental testing, and nine of them have proven to be active at low nanomolar concentrations. One of the confirmed hits, [(8α)-6-methyl-9,10-didehydroergolin-8-yl]methanol), showed very high binding affinity (Ki) of 2.3 nM against 5-HT 1A receptor.

Future Directions and Conclusion

To summarize, we would like to emphasize that QSAR modeling represents a time-, labor-, and cost-effective tool to discover hit compounds and lead candidates in the early stages of drug discovery process. Analyzing the examples of QSAR-based VS available in the literature, one can see that many of them led to the identification of promising lead candidates. However, along with success stories, many QSAR projects fail on the model building stage. This is caused by the lack of understanding that QSAR is highly interdisciplinary and application field as well as general ignorance of the best practices in the field ( Tropsha, 2010 ; Ban et al., 2017 ). Earlier, we have explained this by the undesirably high population of “button pushers,” that is, researchers who conduct modeling without understanding and analyzing the data and modeling process itself ( Muratov et al., 2012 ). This was also explained by the elusive ease of obtaining computational model and making even advanced calculations without understanding of the sense and limitations of the approach ( Bajorath, 2012 ). In addition to this, a lot of even experienced researchers target their efforts to a “vicious statistical cycle,” which main goal is to validate models using as many metrics as possible. In this case, the QSAR modeling is restricted to a single simple question: “What is the best metrics or the best statistical method”? Although we recognize that the right choice of statistical approach and especially rigorous external validation are necessary and represent an essential step in any computer-aided drug discovery study, we want to reinforce that QSAR modeling is useful only if it is applied for the solution of a formulated problem and results in development of new compounds with desired properties.

As future directions, we would like to point out that the era of big data has just started, and it is still in the chemical/biological data accumulation stage. Therefore, to avoid the situation that the number of assayed compounds available on literature exceeds the modeling capability, the development, and implementation of new machine learning algorithms and data curation methods capable of handling millions of compounds are urgently needed. Finally, the overall success of any QSAR-based VS project depends on the ability of a scientist to think critically and prioritize the most promising hits according to his experience. Moreover, the success rate of collaborative drug discovery projects, where the final selection of computational hits is done by both a modeler and an expert in a given field, is much higher than success rate of the projects driven solely by computational or experimental scientists.

Author Contributions

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

This work was partially funded by the Grant No. 1U01CA207160 from NIH and Grant No. 400760/2014-2 from CNPq. CHA is Research Fellow in productivity of CNPq.

Conflict of Interest Statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Acknowledgments

The authors would like to thank Brazilian funding agencies, CNPq, CAPES, and FAPEG, for financial support and fellowships.

AlMatar, M., AlMandeal, H., Var, I., Kayar, B., and Köksal, F. (2017). New drugs for the treatment of Mycobacterium tuberculosis infection. Biomed. Pharmacother. 91, 546–558. doi: 10.1016/j.biopha.2017.04.105

PubMed Abstract | CrossRef Full Text | Google Scholar

Bajorath, J. (2012). Computational chemistry in pharmaceutical research: at the crossroads. J. Comput. Aided. Mol. Des. 26, 11–12. doi: 10.1007/s10822-011-9488-z

Ban, F., Dalal, K., Li, H., LeBlanc, E., Rennie, P. S., and Cherkasov, A. (2017). Best practices of computer-aided drug discovery: lessons learned from the development of a preclinical candidate for prostate cancer with a new mechanism of action. J. Chem. Inf. Model. 57, 1018–1028. doi: 10.1021/acs.jcim.7b00137

Butkiewicz, M., Lowe, E. W., Mueller, R., Mendenhall, J. L., Teixeira, P. L., Weaver, C. D., et al. (2013). Benchmarking ligand-based virtual high-throughput screening with the pubchem database. Molecules 18, 735–756. doi: 10.3390/molecules18010735

Cherkasov, A., Muratov, E. N., Fourches, D., Varnek, A., Baskin, I. I., Cronin, M., et al. (2014). QSAR modeling: where have you been? Where are you going to? J. Med. Chem. 57, 4977–5010. doi: 10.1021/jm4004285

Cihlar, T., and Fordyce, M. (2016). Current status and prospects of HIV treatment. Curr. Opin. Virol. 18, 50–56. doi: 10.1016/j.coviro.2016.03.004

Colley, D. G., Bustinduy, A. L., Secor, W. E., and King, C. H. (2014). Human schistosomiasis. Lancet 383, 2253–2264. doi: 10.1016/S0140-6736(13)61949-2

CrossRef Full Text | Google Scholar

Ekins, S., Lage de Siqueira-Neto, J., McCall, L.-I., Sarker, M., Yadav, M., Ponder, E. L., et al. (2015). Machine learning models and pathway genome data base for Trypanosoma cruzi drug discovery. PLoS Negl. Trop. Dis. 9:e0003878. doi: 10.1371/journal.pntd.0003878

Fourches, D., Muratov, E., and Tropsha, A. (2010). Trust, but verify: on the importance of chemical structure curation in cheminformatics and QSAR modeling research. J. Chem. Inf. Model. 50, 1189–1204. doi: 10.1021/ci100176x

Fourches, D., Muratov, E., and Tropsha, A. (2015). Curation of chemogenomics data. Nat. Chem. Biol. 11, 535–535. doi: 10.1038/nchembio.1881

Fourches, D., Muratov, E., and Tropsha, A. (2016). Trust, but verify II: a practical guide to chemogenomics data curation. J. Chem. Inf. Model. 56, 1243–1252. doi: 10.1021/acs.jcim.6b00129

Garbelli, A., Riva, V., Crespan, E., and Maga, G. (2017). How to win the HIV-1 drug resistance hurdle race: running faster or jumping higher? Biochem. J. 474, 1559–1577. doi: 10.1042/BCJ20160772

Goh, G. B., Hodas, N. O., and Vishnu, A. (2017). Deep learning for computational chemistry. J. Comput. Chem. 38, 1291–1307. doi: 10.1002/jcc.24764

Gomes, M. N. M. N., Braga, R. C. R. C., Grzelak, E. M. E. M., Neves, B. J. B. J., Muratov, E., Ma, R., et al. (2017). QSAR-driven design, synthesis and discovery of potent chalcone derivatives with antitubercular activity. Eur. J. Med. Chem. 137, 126–138. doi: 10.1016/j.ejmech.2017.05.026

Gorobets, N. Y., Sedash, Y. V., Singh, B. K., Poonam, A., and Rathi, B. (2017). An overview of currently available antimalarials. Curr. Top. Med. Chem. 17, 2143–2157. doi: 10.2174/1568026617666170130123520

Hansch, C., and Fujita, T. (1964). p -σ-π analysis. A method for the correlation of biological activity and chemical structure. J. Am. Chem. Soc. 86, 1616–1626. doi: 10.1021/ja01062a035

Kar, S., and Roy, K. (2013). How far can virtual screening take us in drug discovery? Expert Opin. Drug Discov. 8, 245–261. doi: 10.1517/17460441.2013.761204

Kuntz, A. N., Davioud-Charvet, E., Sayed, A. A., Califf, L. L., Dessolin, J., Arnér, E. S. J., et al. (2007). Thioredoxin glutathione reductase from Schistosoma mansoni : an essential parasite enzyme and a key drug target. PLoS Med. 4:e206. doi: 10.1371/journal.pmed.0040206

Kurczyk, A., Warszycki, D., Musiol, R., Kafel, R., Bojarski, A. J., and Polanski, J. (2015). Ligand-based virtual screening in a search for novel anti-HIV-1 chemotypes. J. Chem. Inf. Model. 55, 2168–2177. doi: 10.1021/acs.jcim.5b00295

Laborda, P., Wang, S. Y., and Voglmeir, J. (2016). Influenza neuraminidase inhibitors: synthetic approaches, derivatives and biological activity. Molecules 21, 1–40. doi: 10.3390/molecules21111513

Lacivita, E., Di Pilato, P., De Giorgio, P., Colabufo, N. A., Berardi, F., Perrone, R., et al. (2012). The therapeutic potential of 5-HT1A receptors: a patent review. Expert Opin. Ther. Pat. 22, 887–902. doi: 10.1517/13543776.2012.703654

Lian, W., Fang, J., Li, C., Pang, X., Liu, A.-L., and Du, G.-H. (2015). Discovery of influenza A virus neuraminidase inhibitors using support vector machine and Naïve Bayesian models. Mol. Divers. 20, 439–451. doi: 10.1007/s11030-015-9641-z

Lipinski, C. A., Lombardo, F., Dominy, B. W., and Feeney, P. J. (1997). Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv. Drug Deliv. Rev. 23, 3–25. doi: 10.1016/S0169-409X(96)00423-1

Luo, M., Wang, X. S., Roth, B. L., Golbraikh, A., and Tropsha, A. (2014). Application of quantitative structure-activity relationship models of 5-HT1A receptor binding to virtual screening identifies novel and potent 5-HT1A ligands. J. Chem. Inf. Model. 54, 634–647. doi: 10.1021/ci400460q

Melo-Filho, C. C., Dantas, R. F., Braga, R. C., Neves, B. J., Senger, M. R., Valente, W. C. G., et al. (2016). QSAR-driven discovery of novel chemical scaffolds active against Schistosoma mansoni . J. Chem. Inf. Model. 56, 1357–1372. doi: 10.1021/acs.jcim.6b00055

Menard, D., and Dondorp, A. (2017). Antimalarial drug resistance: a threat to malaria elimination. Cold Spring Harb. Perspect. Med. 7:a025619. doi: 10.1101/cshperspect.a025619

Mitchell, J. B. O. (2014). Machine learning methods in chemoinformatics. Wiley Interdiscip. Rev. Comput. Mol. Sci. 4, 468–481. doi: 10.1002/wcms.1183

Mueller, R., Dawson, E. S., Meiler, J., Rodriguez, A. L., Chauder, B. A., Bates, B. S., et al. (2012). Discovery of 2-(2-Benzoxazoyl amino)-4-Aryl-5-cyanopyrimidine as negative allosteric modulators (NAMs) of metabotropic glutamate receptor5 (mGlu 5): from an artificial neural network virtual screen to an in vivo tool compound. ChemMedChem 7, 406–414. doi: 10.1002/cmdc.201100510

Mueller, R., Rodriguez, A. L., Dawson, E. S., Butkiewicz, M., Nguyen, T. T., Oleszkiewicz, S., et al. (2010). Identification of metabotropic glutamate receptor subtype 5 potentiators using virtual high-throughput screening. ACS Chem. Neurosci. 1, 288–305. doi: 10.1021/cn9000389

Muratov, E. N., Varlamova, E. V., Artemenko, A. G., Polishchuk, P. G., and Kuz’min, V. E. (2012). Existing and developing approaches for QSAR analysis of mixtures. Mol. Inform. 31, 202–221. doi: 10.1002/minf.201100129

Nantasenamat, C., and Prachayasittikul, V. (2015). Maximizing computational tools for successful drug discovery. Expert Opin. Drug Discov. 10, 321–329. doi: 10.1517/17460441.2015.1016497

Neves, B. J., Dantas, R. F., Senger, M. R., Melo-Filho, C. C., Valente, W. C. G., de Almeida, A. C. M., et al. (2016). Discovery of new anti-schistosomal hits by integration of QSAR-based virtual screening and high content screening. J. Med. Chem. 59, 7075–7088. doi: 10.1021/acs.jmedchem.5b02038

Nichols, D. E., and Nichols, C. D. (2008). Serotonin receptors. Chem. Rev. 108, 1614–1641. doi: 10.1021/cr078224o

OECD (2004). OECD Principles for the Validation, for Regulatory Purposes, of (Quantitative) Structure-Activity Relationship Models. Available at: https://www.oecd.org/chemicalsafety/risk-assessment/37849783.pdf [accessed September 20, 2018]

Google Scholar

Phillips, M. A., Burrows, J. N., Manyando, C., van Huijsduijnen, R. H., Van Voorhis, W. C., and Wells, T. N. C. (2017). Malaria. Nat. Rev. Dis. Prim. 3:17050. doi: 10.1038/nrdp.2017.50

Rodriguez, A. L., Grier, M. D., Jones, C. K., Herman, E. J., Kane, A. S., Smith, R. L., et al. (2010). Discovery of novel allosteric modulators of metabotropic glutamate receptor subtype 5 reveals chemical and functional diversity and in vivo activity in rat behavioral models of anxiolytic and antipsychotic activity. Mol. Pharmacol. 78, 1105–1123. doi: 10.1124/mol.110.067207

Southan, C., Várkonyi, P., and Muresan, S. (2009). Quantitative assessment of the expanding complementarity between public and commercial databases of bioactive compounds. J. Cheminform. 1:10. doi: 10.1186/1758-2946-1-10

Tanrikulu, Y., Krüger, B., and Proschak, E. (2013). The holistic integration of virtual screening in drug discovery. Drug Discov. Today 18, 358–364. doi: 10.1016/j.drudis.2013.01.007

Thorne, N., Auld, D. S., and Inglese, J. (2010). Apparent activity in high-throughput screening: origins of compound-dependent assay interference. Curr. Opin. Chem. Biol. 14, 315–324. doi: 10.1016/j.cbpa.2010.03.020

Tropsha, A. (2010). Best practices for QSAR model development, validation, and exploitation. Mol. Inform. 29, 476–488. doi: 10.1002/minf.201000061

WHO (2018a). HIV/AIDS. Available at: http://www.who.int/news-room/fact-sheets/detail/hiv-aids [accessed September 20, 2018].

WHO (2018b). Influenza (Seasonal). Available at: http://www.who.int/news-room/fact-sheets/detail/influenza-(seasonal) [accessed September 20, 2018].

WHO (2018c). Malaria. Available at: http://www.who.int/news-room/fact-sheets/detail/malaria [accessed September 20, 2018].

WHO (2018d). Schistosomiasis. Available at: http://www.who.int/news-room/fact-sheets/detail/schistosomiasis [accessed September 20, 2018].

WHO (2018e). Tuberculosis. Available at: http://www.who.int/news-room/fact-sheets/detail/tuberculosis [accessed September 20, 2018].

Williams, A. J., and Ekins, S. (2011). A quality alert and call for improved curation of public chemistry databases. Drug Discov. Today 16, 747–750. doi: 10.1016/j.drudis.2011.07.007

Young, D., Martin, T., Venkatapathy, R., and Harten, P. (2008). Are the chemical structures in your QSAR correct? QSAR Comb. Sci. 27, 1337–1345. doi: 10.1002/qsar.200810084

Zhang, L., Fourches, D., Sedykh, A., Zhu, H., Golbraikh, A., Ekins, S., et al. (2013). Discovery of novel antimalarial compounds enabled by QSAR-based virtual screening. J. Chem. Inf. Model. 53, 475–492. doi: 10.1021/ci300421n

Keywords : cheminformatics, machine learning, molecular descriptors, computer-assisted drug design, virtual screening

Citation: Neves BJ, Braga RC, Melo-Filho CC, Moreira-Filho JT, Muratov EN and Andrade CH (2018) QSAR-Based Virtual Screening: Advances and Applications in Drug Discovery. Front. Pharmacol. 9:1275. doi: 10.3389/fphar.2018.01275

Received: 11 August 2018; Accepted: 18 October 2018; Published: 13 November 2018.

Reviewed by:

Copyright © 2018 Neves, Braga, Melo-Filho, Moreira-Filho, Muratov and Andrade. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: Carolina Horta Andrade, [email protected] ; [email protected]

Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Published: 01 March 2011

Impact of high-throughput screening in biomedical research

• Ricardo Macarron 1 ,
• Martyn N. Banks 2 ,
• Dejan Bojanic 3 ,
• David J. Burns 4 ,
• Dragan A. Cirovic 5 ,
• Tina Garyantes 6 ,
• Darren V. S. Green 7 ,
• Robert P. Hertzberg 8 ,
• William P. Janzen 9 ,
• Jeff W. Paslay   nAff12 ,
• Ulrich Schopfer 10 &
• G. Sitta Sittampalam   nAff13  

Nature Reviews Drug Discovery volume  10 ,  pages 188–195 ( 2011 ) Cite this article

29k Accesses

857 Citations

299 Altmetric

Metrics details

• Chemical biology
• High-throughput screening
• Medical research

High-throughput screening (HTS) has been postulated in several quarters to be a contributory factor to the decline in productivity in the pharmaceutical industry. Moreover, it has been blamed for stifling the creativity that drug discovery demands. In this article, we aim to dispel these myths and present the case for the use of HTS as part of a proven scientific tool kit, the wider use of which is essential for the discovery of new chemotypes.

This is a preview of subscription content, access via your institution

Access options

Subscribe to this journal

Receive 12 print issues and online access

195,33 € per year

only 16,28 € per issue

Buy this article

• Purchase on SpringerLink
• Instant access to full article PDF

Prices may be subject to local taxes which are calculated during checkout

Similar content being viewed by others

The Pan-Canadian Chemical Library: A Mechanism to Open Academic Chemistry to High-Throughput Virtual Screening

Statistical models for identifying frequent hitters in high throughput screening

Testing the predictive power of reverse screening to infer drug targets, with the help of machine learning

LaMattina, J. L. (ed.) in Drug Truths: Dispelling the Myths About Pharma R&D 114–115 (John Wiley and Sons, Inc., Hoboken, New Jersey, 2009).

Google Scholar  

Graul, A. I., Revel, L., Tell, M., Rosa, E. & Cruces, E. Overcoming the obstacles in the pharma/biotech industry: 2009 update. Drug News Perspect. 23 , 48–63 (2010).

Article   Google Scholar  

Lahana, R. Who wants to be irrational? Drug Discov. Today 8 , 655–656 (2003).

Landers, P. Human Element: drug industry's big push into technology falls short — testing machines were built to streamline research — but may be stifling it — officials see payoff after 2010. The Wall Street Journal 1 (24 Feb 2004).

Ashburn, T. T. & Thor, K. B. Drug repositioning: identifying and developing new uses for existing drugs. Nature Rev. Drug Discov. 3 , 673–683 (2004).

Article   CAS   Google Scholar  

Garnier, J. Rebuilding the R&D engine in big pharma. Harvard Bus. Rev. 86 , 68–76 (2008).

Zhang, J. H., Chung, T. D. & Oldenburg, K. R. A simple statistical parameter for use in evaluation and validation of high throughput screening assays. J. Biomol. Screen. 4 , 67–73 (1999).

Makarenkov, V. et al. An efficient method for the detection and elimination of systematic error in high-throughput screening. Bioinformatics 23 , 1648–1657 (2007).

Coma, I. et al. Process validation and screen reproducibility in high-throughput screening. J. Biomol. Screen. 14 , 66–76 (2009).

Taylor, P. B. et al. A standard operation procedure for assessing liquid handler performance in high-throughput screening. J. Biomol. Screen. 7 , 554–569 (2002).

Paul, S. M. et al. How to improve R&D productivity: the pharmaceutical industry's grand challenge. Nature Rev. Drug Discov. 9 , 203–214 (2010).

CAS   Google Scholar  

Fox, S. et al. High-throughput screening: update on practices and success. J. Biomol. Screen. 11 , 864–869 (2006).

Bleicher, K. H., Bohm, H.-J., Muller, K. & Alanine, A. I. Hit and lead generation: beyond high-throughput screening, Nature Rev. Drug Discov. 2 , 369–378 (2003).

Mullin, R. As high-throughput screening draws fire, researchers leverage science to put automation into perspective. Chem. Eng. News 82 , 23–32 (2004).

Kola, I. & Landis, J. Can the pharmaceutical industry reduce attrition rates? Nature Rev. Drug Discov. 3 , 711–715 (2004).

Perola, E. An analysis of the binding efficiencies of drugs and their leads in successful drug discovery programs. J. Med. Chem. 53 , 2986–2997 (2010).

Dorr, P. et al. Maraviroc (UK-427,857), a potent, orally bioavailable, and selective small-molecule inhibitor of chemokine receptor CCR5 with broad-spectrum anti-human immunodeficiency virus type 1 activity. Antimicrob. Agents Chemother. 49 , 4721–4732 (2005).

Duffy, K. J. et al. Hydrazinonaphthalene and azonaphthalene thrombopoietin mimics are nonpeptidyl promoters of megakaryocytopoiesis. J. Med. Chem. 44 , 3730–3745 (2001).

Duffy, K. J. et al. Identification of a pharmacophore for thrombopoietic activity of small, non-peptidyl molecules. 1. Discovery and optimization of salicylaldehyde thiosemicarbazone thrombopoietin mimics. J. Med. Chem. 45 , 3573–3575 (2002).

Duffy, K. J. et al. Identification of a pharmacophore for thrombopoietic activity of small, non-peptidyl molecules. 2. Rational design of naphtho[1,2-d]imidazole thrombopoietin mimics. J. Med. Chem. 45 , 3576–3578 (2002).

Erickson-Miller, C. L. et al. Discovery and characterization of a selective, nonpeptidyl thrombopoietin receptor agonist. Exp. Hematol. 33 , 85–93 (2005).

Cuatrecasas, P. Drug discovery in jeopardy. J. Clin. Invest. 116 , 2837–2842 (2006).

Lemm, J. A. et al. Identification of hepatitis C virus NS5A inhibitors. J. Virol. 84 , 482–491 (2010).

Gao, M. et al. Chemical genetics strategy identifies an HCV NS5A inhibitor with a potent clinical effect. Nature 456 , 96–100 (2010).

Keserü, G. M. & Makara, G. M. The influence of lead discovery strategies on the properties of drug candidates. Nature Rev. Drug Discov. 8 , 203–212 (2009).

Oprea, T. I., Davis, A. M., Teague, S. J. & Leeson, P. D. Is there a difference between leads and drugs? A historical perspective. J. Chem. Inf. Comput. Sci. 41 , 1308–1315 (2001).

Hann, M. M. & Oprea, T. I. Pursuing the leadlikeness concept in pharmaceutical research. Curr. Opin. Chem. Biol. 8 , 255–263 (2004).

Jacoby, E. et al. Key aspects of the Novartis compound collection enhancement project for the compilation of a comprehensive chemogenomics drug discovery screening collection. Curr. Top. Med. Chem. 5 , 397–411 (2005).

Lane, S. J. et al. Defining and maintaining a high quality screening collection: the GSK experience. Drug Discov. Today 11 , 267–272 (2006).

HTStec Limited. Cellular Assay Reagents Trends 2009 Report (Cambridge, UK, 2009).

Drewry, D. H. & Macarron, R. Enhancements of screening collections to address areas of unmet medical need: an industry perspective. Curr. Opin. Chem. Biol. 14 , 289–298 (2010).

Fink, T. & Reymond, J.-L. Virtual exploration of the chemical universe up to 11 atoms of C, N, O, F: assembly of 26.4 million structures (110.9 million stereoisomers) and analysis for new ring systems, stereochemistry, physicochemical properties, compound classes, and drug discovery. J. Chem. Inf. Model. 47 , 342–353 (2007).

Harper, G., Pickett, S. D. & Green, D. V. S. Design of a compound screening collection for use in high throughput screening. Comb. Chem. High Throughput Screen. 7 , 63–70 (2004).

Martin, Y. C., Kofron, J. L. & Traphagen, L. M. Do structurally similar molecules have similar biological activity? J. Med. Chem. 45 , 4350–4358 (2002).

Munos, B. Lessons from 60 years of pharmaceutical innovation. Nature Rev. Drug Discov. 8 , 959–968 (2009).

Austin, C. P. The completed human genome: implications in chemical biology. Curr. Opin. Chem. Biol. 7 , 511–515 (2003).

Society for Laboratory Automation and Screening. Screening facilities. SLAS [online], http://www.slas.org/screeningFacilities/facilityList.cfm , (2010).

Molecular Libraries Program. MLPCN probes Web table. MLP [online], http://mli.nih.gov/mli/mlp-probes/ , (2010).

Kaiser, J. Industrial-style screening meets academic biology. Science 321 , 764–766 (2008).

Silber, B. M. Driving drug discovery: the fundamental role of academic labs. Sci. Transl. Med. 2 , 30cm16 (2010).

Frye, S. V. The art of the chemical probe. Nature Chem. Biol. 6 , 159–161 (2010).

Houston, J. G. et al. Case study: impact of technology investment on lead discovery at Bristol-Myers Squibb, 1998–2006. Drug Discov. Today 13 , 44–51 (2008).

Banks, M. N., Zhang, L. & Houston, J. G. in Exploiting Chemical Diversity for Drug Discovery (eds Bartlett, P. A. & Entzeroth, M.) 315–335 (Royal Chemical Society Publishing, London, 2006).

Kunapuli, P. et al. Application of division arrest technology to cell-based HTS: comparison with frozen and fresh cells. Assay Drug Dev. Technol. 3 , 17–26 (2005).

Digan, M. E., Pou, C., Niu, H. & Zhang, J. H. Evaluation of division-arrested cells for cell-based high-throughput screening and profiling. J. Biomol. Screen. 10 , 615–623 (2005).

Cawkill, D. & Eaglestone S. S. Evolution of cell-based reagent provision. Drug Discov. Today 12 , 820–825 (2007).

Zaman, G. J. et al. Cryopreserved cells facilitate cell-based drug discovery. Drug Discov. Today 12 , 521–526 (2007).

Leifert, W. R., Aloia, A. L., Bucco, O., Glatz, R. V. & McMurchie, E. J. G-protein coupled receptors in drug-discovery: nano-sizing using cell-free technologies and molecular biology approaches. J. Biomol. Screen. 10 , 765–779 (2005).

Heilker, R., Zemanova, L., Valler, M. J. & Niehaus, G. U. Confocal fluorescence microscopy for high-throughput screening of G-protein coupled receptors. Curr. Med. Chem. 12 , 2551–2255 (2005).

Hovius, R., Vallotton, P., Wohland, T. & Vogel, H. Fluorescence techniques: shedding light on ligand-receptor interactions. Trends Pharmacol. Sci. 21 , 266–273 (2000).

Jia, Y., Quinn, C. M., Kwak, S. & Talanian, R. V. Current in vitro kinase assay technologies: the quest for a universal format. Curr. Drug Discov. Technol. 5 , 59–69 (2008).

Gasparri, F., Sola, F., Bandiera, T., Moll, J. & Galvani, A. High content analysis of kinase activity in cells. Comb. Chem. High Throughput Screen. 11 , 523–536 (2008).

McLoughlin, D. J., Bertelli, F. & Williams, C. The A, B, Cs of G-protein-coupled receptor pharmacology in assay development for HTS. Expert Opin. Drug Discov. 2 , 1–17 (2007).

Williams, C. & Sewing, A. G-protein-coupled receptor assays: to measure affinity or efficacy that is the question. Comb. Chem. High Throughput Screen. 8 , 285–292 (2005).

Wunder, F., Kalthof, B., Müller, T. & Hüser, J. Functional cell-based assays in microliter volumes for ultra-high throughput screening. Comb. Chem. High Throughput Screen. 11 , 495–504 (2008).

Kenakin, T. P. Cellular assays as portals to seven-transmembrane receptor-based drug discovery. Nature Rev. Drug Discov. 8 , 617–626 (2009).

Fry, D. W. et al. A specific inhibitor of the epidermal growth factor receptor tyrosine kinase. Science 265 , 1093–1095 (1994).

Wilhelm, S. et al. Discovery and development of sorafenib: a multikinase inhibitor for treating cancer. Nature Rev. Drug Discov. 5 , 835–844 (2006).

Thaisrivongs, S. et al. Structure-based design of HIV protease inhibitors: 4-hydroxycoumarins and 4-hydroxy-2-pyrones as nonpeptidic inhibitors. J. Med. Chem. 37 , 3200–3204 (1994).

Thornberry, N. A. & Weber, A. E. Discovery of JANUVIA ™ (Sitagliptin), a selective dipeptidyl peptidase IV inhibitor for the treatment of type 2 diabetes. Curr. Top. Med. Chem. 7 , 557–568 (2007).

Das, J. et al. 2-aminothiazole as a novel kinase inhibitor template. Structure–activity relationship studies toward the discovery of N -(2-chloro-6-methylphenyl)-2-[[6-[4-(2-hydroxyethyl)- 1-piperazinyl)]-2-methyl-4- pyrimidinyl]amino)]-1, 3-thiazole-5-carboxamide (Dasatinib, BMS-354825) as a potent pan -Src kinase inhibitor. J. Med. Chem. 49 , 6819–6832 (2006).

LaMattina, J. L. (ed.) in Drug Truths: Dispelling the Myths About Pharma R&D 65–66 (John Wiley and Sons, Inc., Hoboken, New Jersey, 2009).

Riechers, H. et al. Discovery and optimization of a novel class of orally active nonpeptidic endothelin-A receptor antagonists. J. Med. Chem. 39 , 2123–2128 (1996).

De Corte, B. L. From 4,5,6,7-tetrahydro-5-methylimidazo[4,5,1-jk](1,4)benzodiazepin-2(1 H )-one (TIBO) to etravirine (TMC125): fifteen years of research on non-nucleoside inhibitors of HIV-1 reverse transcriptase. J. Med. Chem. 48 , 1689–1696 (2005).

Yamamura, Y. et al. OPC-21268, an orally effective, nonpeptide vasopressin VI receptor antagonist. Science 252 , 572–574 (1991).

Inglese, J. in Ask the Expert Forum, ACS Chemical Biology. ACS Publications [online] http://community.acs.org/ChemBiol/AsktheExpert/AsktheExpertArchive/tabid/67/Default.aspx (2008).

Download references

Acknowledgements

The authors are grateful to the many colleagues in the HTS community who have contributed data and opinions presented in this article.

Author information

Jeff W. Paslay

Present address: G.Sitta Sittampalam is at the Department of Pharmacology, Toxicology & Therapeutics, Institute for Advancing Medical Innovations, The University of Kansas Cancer Center, 3901 Rainbow Blvd., Kansas City, Kansas 66160, USA,

G. Sitta Sittampalam

Present address: Jeff W.Paslay was previously at the Department of Screening Sciences, Research Centers of Emphasis, Pfizer; Present address: 302 Old Barn Circle, Phoenixville, Pennsylvania 19460, USA,

Authors and Affiliations

Ricardo Macarron is at the Department of Sample Management Technologies, GlaxoSmithKline, Mail Stop UP 12-200, 1250S Collegeville Rd, Collegeville, Pennsylvania 19426, USA,

Ricardo Macarron

Martyn N. Banks is at the Department of Applied Biotechnologies, Bristol-Myers Squibb Co., 5 Research Parkway, Wallingford, Connecticut 06492, USA,

Martyn N. Banks

Dejan Bojanic is at the Lead Finding Platform, Novartis Institutes for Biomedical Research Inc., 250 Massachusetts Avenue, Cambridge, Massachusetts 02139, USA,

Dejan Bojanic

David J.Burns is with the Early Pain Discovery Team, Abbott Laboratories, R4DI AP52N 200 Abbott Park Road, Abbott Park, Illinois 60064, USA,

David J. Burns

Dragan A.Cirovic is at the Department of Lead Generation and Candidate Realization, Sanofi-Aventis, 1041 Route 202-206, Mail Stop BRJR1-002A, Bridgewater, New Jersey 08807, USA,

Dragan A. Cirovic

Tina Garyantes is at the Department of Lead Generation and Candidate Realization, Sanofi-Aventis, 1041 Route 202-206, P.O. BOX 6800, JR-303D, Bridgewater, New Jersey 08807, USA,

Tina Garyantes

Darren V.S.Green is at the Department of Computational and Structural Chemistry, GlaxoSmithKline, Stevenage, Hertfordshire SG1 2NY, UK,

Darren V. S. Green

Robert P.Hertzberg is at the Department of Screening and Compound Profiling, GlaxoSmithKline, Mail Stop UP 12-L05, 1250 S Collegeville Rd, Collegeville, Pennsylvania 19426, USA,

Robert P. Hertzberg

William P.Janzen is at the Division of Medicinal Chemistry and Natural Products, Center for Integrative Chemical Biology and Drug Discovery, Eshelman School of Pharmacy, The University of North Carolina, Chapel Hill, North Carolina 27599-7363, USA,

William P. Janzen

Ulrich Schopfer is at the Lead Finding Platform, Novartis Institutes for BioMedical Research, Forum 1, Novartis Campus, CH-4056 Basel, Switzerland,

Ulrich Schopfer

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Ricardo Macarron .

Ethics declarations

Competing interests.

The authors declare no competing financial interests.

Supplementary information

Supplementary information table s1.

Molecular Libraries HTS Project Success- Key Indicators and Metrics 2004-2009* (PDF 242 kb)

A popular reporter gene assay that uses a luciferase gene to detect metabolites (for example, cyclic AMP levels) or changes in expression of a gene of interest.

The space spanned by all energetically stable stoichiometric combinations of electrons, atomic nuclei and topologies in molecules. It is calculated to contain up to 1 × 10 60 distinct molecules. Drug-like space may contain up to 1 × 10 30 molecules.

The calculated logarithm of the partition coefficient between n -octanol and water for a given compound. This parameter is an estimation of the lipophilicity of the compound.

Rapid and parallel synthesis of large collections of compounds to facilitate the identification of new active compounds for drug targets by high-throughput screening techniques.

The process of finding the most favourable condition that satisfies all conditions (or constraints) that frame the problem.

Sharing certain characteristics — such as size, shape and solubility in water and organic solvents — with other molecules that act in the same way as drugs. Lipinski's rule of five provides a commonly used definition of these properties for oral drugs.

Computational models designed to predict the ADMET (absorption, distribution, metabolism, excretion and toxicity) of molecules.

The identification of bioactive substances by screening small-molecule fragments (<300 Da). It requires high-resolution structural techniques to guide the optimization of weak efficient hits into leads.

Sharing certain characteristics — such as size, shape and solubility in water and organic solvents — with other molecules that act as precursors of drugs (leads). Lead-likeness is typically associated with small size (molecular mass <400 Da) and low lipophilicity (clogP <4).

Microtitre plates may suffer from heterogeneous temperature, air flow, reader and liquid handler bias, and so on, leading to systematic assay errors that need to be detected and corrected by ad hoc algorithms.

A screen based on whole cells that measures an observable change in cell physiology or morphology in the presence of active compounds. Phenotypic assays cannot distinguish direct compound interactions with the specific targets or signalling pathways in the cell.

Correlations that are constructed between the features of chemical structures in a set of candidate compounds and parameters of biological activity, such as potency, selectivity and toxicity.

The use of three-dimensional structural information and molecular-modelling techniques to design a series of possible pharmacological modulators that can, for example, block an active site of an enzyme.

The selection of chemical compounds that are related to either known ligands of a target or to the target class of interest.

The selection of potential bioactive substances from a much larger list of candidate molecules using in silico models typically based on known structures and/or ligands of the target of interest.

Z′ is a relative indication of the separation of the signal and background controls and is widely used in high-throughput screening (HTS) to assess the quality of an assay. Every microtitre plate in a run will exhibit a distinct Z′ value and monitoring its trends in a campaign is a standard quality control practice.

Rights and permissions

Reprints and permissions

About this article

Cite this article.

Macarron, R., Banks, M., Bojanic, D. et al. Impact of high-throughput screening in biomedical research. Nat Rev Drug Discov 10 , 188–195 (2011). https://doi.org/10.1038/nrd3368

Download citation

Published : 01 March 2011

Issue Date : March 2011

DOI : https://doi.org/10.1038/nrd3368

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

This article is cited by

Qhtswaterfall: 3-dimensional visualization software for quantitative high-throughput screening (qhts) data.

• Bryan Queme
• John C. Braisted
• James Inglese

Journal of Cheminformatics (2023)

PMF-CPI: assessing drug selectivity with a pretrained multi-functional model for compound–protein interactions

• Ruihan Dong

A high-throughput platform for efficient exploration of functional polypeptide chemical space

• Haisen Zhou

Nature Synthesis (2023)

Modeling the expansion of virtual screening libraries

• Jiankun Lyu
• John J. Irwin
• Brian K. Shoichet

Nature Chemical Biology (2023)

CLOOME: contrastive learning unlocks bioimaging databases for queries with chemical structures

• Ana Sanchez-Fernandez
• Elisabeth Rumetshofer
• Günter Klambauer

Nature Communications (2023)

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

• Open access
• Published: 26 March 2023

A systematic review of substance use screening in outpatient behavioral health settings

• Diana Woodward 1 ,
• Timothy E. Wilens 1 ,
• Meyer Glantz 2 ,
• Vinod Rao 1 ,
• Colin Burke 1 &
• Amy M. Yule   ORCID: orcid.org/0000-0002-2409-9426 3  

Addiction Science & Clinical Practice volume  18 , Article number:  18 ( 2023 ) Cite this article

3188 Accesses

3 Citations

9 Altmetric

Metrics details

Despite the frequent comorbidity of substance use disorders (SUDs) and psychiatric disorders, it remains unclear if screening for substance use in behavioral health clinics is a common practice. The aim of this review is to examine what is known about systematic screening for substance use in outpatient behavioral health clinics.

We conducted a PRISMA-based systematic literature search assessing substance use screening in outpatient adult and pediatric behavioral health settings in PubMed, Embase, and PsycINFO. Quantitative studies published in English before May 22, 2020 that reported the percentage of patients who completed screening were included.

Only eight articles met our inclusion and exclusion criteria. Reported prevalence of screening ranged from 48 to 100%, with half of the studies successfully screening more than 75% of their patient population. There were limited data on patient demographics for individuals who were and were not screened (e.g., gender, race) and screening practices (e.g., electronic versus paper/pencil administration).

Conclusions

The results of this systematic review suggest that successful screening for substance use in behavioral health settings is possible, yet it remains unclear how frequently screening occurs. Given the high rates of comorbid SUD and psychopathology, future research is necessary regarding patient and clinic-level variables that may impact the successful implementation of substance use screening.

Trial registry A methodological protocol was registered with the PROSPERO systematic review protocol registry (ID: CRD42020188645).

Introduction

Substance use disorders (SUD) pose a substantial societal burden in the United States. In 2020 alone, an estimated 28.3 million people aged 12 or older met criteria for a past-year alcohol use disorder, while 18.4 million people aged 12 or older experienced a past-year illicit drug use disorder [ 1 ] Risky substance use and SUD are associated with substantial disability and mortality, with an estimated 480000 tobacco-related deaths and 95000 alcohol-related deaths annually in the United States [ 2 , 3 ]. Of particular concern, drug-related overdose deaths have risen over the past years, increasing from 70,630 deaths in 2019 to 92000 deaths in 2020 [ 4 , 5 ].

Prior research has established psychopathology as a significant risk factor for developing a SUD [ 6 , 7 , 8 , 9 ]. For example, individuals with depression are approximately 2 times more likely to develop a SUD, and those with attention deficit hyperactivity disorder exhibit a 2.3 times greater risk [ 10 ]. Furthermore, individuals with one or more psychiatric diagnoses experience greater SUD severity [ 11 , 12 ]. The sequelae of co-occurring SUD and psychiatric disorders include increased odds of additional psychopathology [ 15 ], hospitalizations [ 16 ], suicide attempts [ 13 , 17 , 18 ], overdose [ 19 , 20 , 21 ], criminal behavior [ 22 ], and homelessness [ 23 ]. Additionally, adults with co-occurring disorders report overall lower quality of life [ 24 ] and lower social and occupational functioning [ 13 , 25 , 26 ].

Despite the imposed burden of comorbid SUD and psychopathology, in 2019, 51.4% of individuals in the United States with co-occurring disorders received no treatment, 38.7% received mental health treatment only, 7.8% received treatment for both mental health and SUD, and 1.9% received SUD treatment only (27). Given that many treatment-seeking individuals with co-occurring SUD and psychopathology obtain mental health treatment rather than substance use treatment, screening for substance use concerns in behavioral health settings is necessary to identify individuals at the greatest risk for maladaptive outcomes.

To this end, both the Substance Abuse and Mental Health Services Administration (SAMHSA) and the National Institute for Health and Clinical Guidance (NICE) have urged mental health providers to routinely administer patient self-report questionnaires to screen for substance use [ 28 , 29 ]. Most efforts to integrate substance use screening into clinical care have focused on primary care settings [ 30 , 31 , 32 , 33 ]. As such, the success of substance use screening tools in other outpatient settings remains unclear. Because behavioral health clinics generally have both fewer ancillary supports to assist with screening compared to primary care, as well as high staff turnover rates [ 34 , 35 ], research is needed on screening for substance use in these settings. Hence, we aim to summarize the extant literature on systematic screening for substance use in behavioral health, with a focus on the prevalence of screening within these clinics, characteristics of the screening tools used, and screening practices.

A methodological protocol was registered with the PROSPERO systematic review protocol registry (ID: CRD42020188645).

Search strategy

We conducted a search based upon Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines of peer-reviewed literature within the PubMed, Embase, and PsycINFO databases through May 22, 2020 with no restrictions for the start date. We examined both the prevalence and frequency of substance use screening in outpatient behavioral health clinics as well as the characteristics of the outpatient behavioral health clinics that screen for substance use. We searched each database using various combinations of search terms that can be found in the Additional file 1 . Bibliographies of reviewed articles were also examined for additional studies to ensure that no relevant articles were omitted.

Inclusion criteria were quantitative studies examining substance use screening in outpatient adult and pediatric behavioral health clinics published in English. This included general psychiatric clinics, community mental health organizations, university counseling centers, and other specialty services. Studies were only included if they implemented systematic screening for substance use and reported the percentage of patients who completed the screener. Editorials, commentaries, opinion papers, chapters, and research studies that recruited participants to complete screening tools were excluded. Studies examining screening for substance use only in integrated behavioral health settings within primary care, emergency rooms, or inpatient settings were also excluded. If studies examined screening for substance use in behavioral health-only clinics alongside integrated behavioral health settings they were included if they stratified screening rates by clinic type.

Selection of studies

Two reviewers independently screened the titles and abstracts of all papers. Any disagreements were resolved by consensus, and irrelevant titles were excluded. A record was kept of all irrelevant and duplicate articles. The full text of the remaining papers was reviewed by the two investigators and included/excluded. A third senior investigator reviewed all the included papers to confirm they met inclusion/exclusion criteria.

Data extraction, analysis, and synthesis

Data were extracted from the quantitative studies by one reviewer and discussed with the senior reviewer. The following variables were extracted: Setting, sample size, percentage of patients screened, patient demographics, language of screening tool, screener administered, substances screened, date of study, frequency of screening, and method of screening (computer, paper, self-report, clinician report, etc.).

Our initial search yielded 362 non-duplicate articles (Fig.  1 ). Eighty-four articles were determined to be potentially relevant and therefore reviewed in full. Of the 84 potentially relevant articles, 76 articles were excluded based on eligibility criteria (see Fig.  1 ). Eight articles were included in the final review (Table 1 ). The most common reasons for exclusion after full-text review were that the article reported on data from a sample recruited for a research study (N = 23), the authors did not report the percentage of patients screened (N = 13), or screening was implemented in a non-behavioral health setting (N = 11). The 8 articles included in this systematic review were published between 1992 and 2018. The sample sizes ranged from 88 to 22,956 screened patients.

PRISMA diagram

Six of the eight studies were conducted in behavioral health clinics within a larger healthcare system, two of which took place in Veterans Affairs (VA) facilities [ 36 , 37 , 38 , 39 , 40 , 41 ]. The two studies that were not conducted in healthcare systems were conducted in a university counseling center [ 42 ] and community mental health organizations [ 43 ]. All studies were conducted in the United States. Four studies were single-site [ 37 , 40 , 41 , 42 ], three studies included multiple sites ranging from 2 to 48 [ 36 , 39 , 43 ], and one study did not report the number of sites [ 38 ]. The majority of the studies (62.5%) were conducted in adult clinics [ 37 , 39 , 40 , 42 , 44 ], with one study focused on college students [ 42 ]. Two studies included pediatric patients [ 36 , 43 ], and one study did not report age [ 41 ].

Screener and substances screened

All of the studies screened for alcohol, the majority screened for drugs (N = 6) [ 36 , 37 , 39 , 40 , 42 , 43 ], and half of the studies screened for tobacco (N = 4) [ 36 , 39 , 40 , 42 ]. Of those that screened for drugs, two studies administered screeners which did not differentiate type of substance [ 36 , 37 ]. Of the remaining four, all specifically queried about marijuana/cannabis [ 39 , 40 , 42 , 43 ], and three screened for other drugs, including opioids [ 39 , 40 , 42 ]. A range of 1 to 5 screeners was used to assess for substance use. Additionally, one study administered both a pre-screening instrument and a screening instrument [ 42 ]. The most commonly used screeners were the Alcohol Use Disorders Identification Test-Concise (AUDIT-C) [ 38 , 42 , 45 ], the CRAFFT [ 36 , 43 , 46 ], the Alcohol, Smoking, and Substance Involvement Screening Test (ASSIST) [ 39 , 42 , 47 ], and the Short Michigan Alcohol Screening Test (SMAST) [ 40 , 41 , 48 ] (all N = 2).

Frequency and methods of screening

The majority of studies reported screening only at intake (N = 6) [ 37 , 38 , 39 , 40 , 41 , 42 ]. One clinic implemented different screening instruments at intake, quarterly, and one year [ 36 ], and Stanhope et al. did not report the frequency of their screening across community mental health organizations. Of the eight studies, five relied solely on self-administration [ 36 , 37 , 39 , 40 , 41 ], one on both self- (prescreen) and clinician- (screen) administration [ 42 ], and two did not report how the screening was administered [ 38 , 43 ]. Additionally, although the majority (N = 5) of authors did not report how information was collected [ 36 , 38 , 41 , 42 , 43 ], two studies utilized an electronic screen [ 39 , 40 ] and one study relied on paper and pencil [ 37 ]. Finally, none of the studies reported the language of their screening instrument(s) [ 37 ].

Screening rate

One study reported screening all patients [ 42 ]. The screening rates of the remaining studies ranged from 48 to 93.5% of patients. Screening in adult-only clinics ranged from 48 to 100% of patients [ 37 , 38 , 39 , 40 , 42 ] while screening from clinics with adult and pediatric patients ranged from 84 to 93.5% [ 36 , 43 ]. The screening rate using an electronic screen ranged from 48 to 75% of patients [ 39 , 40 ], and the rate for paper/pencil was 74.9%.

Demographics

Five studies reported on the gender of screened patients [ 36 , 37 , 39 , 40 , 43 ], and one study reported on gender across the total study population (patients who did and did not complete the screening) [ 38 ]. Of those that reported on the gender of screened patients, the range was 30 to 86% male. In two studies that did not report the gender across the total study population, the studies did report that there were no significant gender differences between patients who did and did not complete screening [ 40 , 43 ].

Four studies reported the mean age of screened patients, with a range of 16.6 to 42.9 years [ 37 , 39 , 40 , 43 ]. Three studies reported mean age across the total study population, with a range of 36.1 to 53.5 years [ 38 , 39 , 40 ]. One additional study that did not report mean age across the total study population reported no significant difference in mean age between screened patients and the total study population [ 43 ].

The three studies that reported on the race of screened patients included predominantly white patients, with these participants ranging from 52.8 to 72% of the sample [ 39 , 40 , 43 ]. The next most represented race was Asian, ranging from 9 to 10.5% of the sample. No studies reported race across the total study population; however, two studies reported no significant racial differences between patients who did and did not complete screening [ 40 , 43 ].

The two studies that reported on the ethnicity of screened patients included predominantly non-Hispanic patients, with these patients ranging from 73 to 93% of the sample [ 40 , 43 ]. The one study that reported ethnicity across the total study population (patients who did and did not complete screening) was also largely non-Hispanic (94.2%) [ 38 ]. Two additional studies reported no significant differences in ethnicity between patients who did and did not complete screening; however, they did not report ethnicity type for the study population [ 40 , 43 ].

Psychiatric comorbidities

Though two studies provided descriptive information on psychopathology, neither compared psychopathology between those who were and were not screened in the clinic [ 37 , 38 ]. Karno et al. reported rates of depressive disorder (48%), anxiety disorder (15%), bipolar disorder (13%), and schizophrenia/ schizoaffective disorder (11%) in screened patients. King and colleagues found that 15.1% of all clinic patients had a trauma/ stressor-related disorder (including post-traumatic stress disorder), and 12.9% of all clinic patients had a mood disorder.

Our aim in this review was to determine the prevalence and the characteristics of screening practices for substance use in outpatient behavioral health clinics. Though we identified only 8 studies that met review criteria, half of these studies reported screening more than 75% of their patient population [ 36 , 41 , 42 , 43 ].

The screening rates in the identified studies are comparable to those reported in a recent examination of substance use screening in primary care settings, which found that 71.8% of eligible patients were screened after implementation efforts [ 49 ]. However, whether existing research on screening for substance use represents standard practice in all behavioral health clinics remains unclear given limited reporting on this practice. While the 2020 National Mental Health Services Survey (N-MHSS) reported that approximately 54% of the 4,941 surveyed outpatient mental health treatment facilitates offer screening for tobacco use, it did not specify whether this screening is systematic and routine and did not report on screening for non-nicotine substances (50). Furthermore, the intent to screen for substance use does not always translate into clinical practice. A large survey found that although 93.1% and 78.9% of mental health clinic directors reported having screening guidelines for alcohol and illicit substance use, respectively, only 66.6% and 57.8% of clinic staff reported conducting said screening [ 51 ].

Several patient- and clinic-level variables influence the successful implementation of systematic screening. Unfortunately, few studies in the current review reported patient demographic information. We were therefore unable to identify specific patient demographics associated with a high prevalence of screening for substance use or demographic differences between patients who were and were not screened to help identify patient groups who did not complete screening. This is notable since research from the primary care setting has found differences in screening for substance use based on demographics. For example, Black and Hispanic patients and adults over the age of 65 may require more assistance to complete electronic screening for substance use due to problems with comprehension or technical issues [ 52 ]. In light of increasing overdose deaths among Black and Hispanic youth [ 53 ], research examining the barriers to screening for substance use in particular demographic groups is needed to ensure equitable care.

Clinic factors that influence the successful implementation of screening center around the method of screening administration. For most studies in our review, screening tools were administered as patient self-report [ 36 , 37 , 39 , 40 , 41 ]. This is consistent with recent research in primary care and emergency department settings showing increased patient comfort with self-report screening compared to clinician-administered screening [ 54 , 55 , 56 ], particularly amongst individuals who belong to groups who are more stigmatized for substance use [ 52 , 57 , 58 ]. Another notable finding of our review was the omission of data regarding screening tools (paper and pencil versus electronic) and language of screening. A review of screening in primary care found that electronic questionnaires using patient self-report in both pediatric and adult settings improved data quality and completion time, decreased costs, and were preferred by patients. However, the use of electronic questionnaires also led to increased privacy concerns and access challenges [ 59 ]. Electronic measures, particularly those linked to the electronic medical record, may also result in racial and ethnic disparities in screening completion rates [ 60 ]. Additional research in the behavioral health setting is needed to determine patient and clinician preferences regarding the method of screening, particularly for more stigmatized conditions such as substance use [ 61 , 62 ].

Finally, the timing and frequency of screening is another important factor to consider during implementation. Most studies in our systematic review reported screening patients for substance use only at intake [ 36 , 37 , 38 , 39 , 40 , 41 , 42 , 43 ]. Although screening at intake identifies patients who may benefit from SUD treatment [ 63 ], ongoing screening and progress monitoring improves engagement in SUD treatment and SUD outcomes [ 64 , 65 ], and a recent consensus panel organized by SAMHSA recommended screening patients with psychiatric disorders for substance use annually [ 66 ]. Thus, future research should examine the prevalence and success of repeated screening for substance use.

The results of our review need to be considered in light of methodological limitations. The generalizability of the findings may be limited given the small number of eligible manuscripts. Moreover, several of these studies were missing information on patient- and clinic-level variables related to implementation that was recently identified as necessary to report on for studies evaluating the use of patient self-report questionnaires to improve the methodological quality, transparency, and applicability of the findings [ 67 ]. Hence it was difficult to conclude what variables contributed to the successful implementation of screening for substance use in the behavioral health setting. Furthermore, of those studies that did report patient demographics, the majority of the subjects were adults, white, and non-Hispanic. As such, the results may not be generalizable to pediatric or more diverse racial and ethnic groups. Additionally, to narrow the scope of the current review, we excluded manuscripts that examined substance use screening in integrated behavioral health clinics within primary care. Although implementation in these settings is important to investigate to better understand the overall landscape of screening for substance use in settings that provide behavioral health care, integrated behavioral health clinics likely face different barriers and facilitators. Lastly, more clinics may be systematically screening for substance use and not reporting their findings in published results. Thus, this topic is at risk for publication bias as behavioral health clinics that have struggled to implement systematic screening for substance use may not pursue publication.

In summary, the results of our review indicate that screening for substance use in the outpatient behavioral health setting can be successfully implemented at initial intake. Our review highlights the need for further examination of patient- and clinic-level variables that may impact the successful implementation of screening in behavioral health. Future research should include these variables to inform implementation efforts, ensure equity in screening, and achieve consistency with recent reporting guidelines [ 67 ].

Availability of data and materials

All data generated or analyzed during this study are included in this published article [and its Additional information files].

Abbreviations

• Substance use disorder

Substance Abuse and Mental Health Services Administration

National Institute for Health and Clinical Guidance

Preferred reporting items for systematic reviews and meta-analyses

Veterans affairs

Alcohol use disorders identification test

Car, relax, alone, forget, friends, trouble

Alcohol, smoking and substance involvement screening test

Short Michigan alcohol screening test

National Mental Health Services Survey

Substance Abuse and Mental Health Services Administration. Key Substance Use and Mental Health Indicators in the United States: Results from the 2020 National Survey on Drug Use and Health. 2020;156. https://www.samhsa.gov/data/sites/default/files/reports/rpt35325/NSDUHFFRPDFWHTMLFiles2020/2020NSDUHFFR1PDFW102121.pdf . Accessed 14 July 2022.

Esser MB, Sherk A, Liu Y, Naimi TS, Stockwell T, Stahre M, et al. Deaths and years of potential life lost from excessive alcohol use—United States, 2011–2015. MMWR Morb Mortal Wkly Rep. 2020;69(39):1428–33. https://doi.org/10.15585/mmwr.mm6939a6 .

Article   PubMed   PubMed Central   Google Scholar  

National Center for Chronic Disease Prevention and Health Promotion (US) Office on Smoking and Health. The health consequences of smoking—50 years of progress: a report of the surgeon general. centers for disease control and prevention (US). 2014. https://www.ncbi.nlm.nih.gov/books/NBK179276/ . Accessed 2 Aug. 2022.

Ahmad FB, Cisewski JA, Rossen LM, Sutton P. Provisional drug overdose death counts national center for health statistics. 2022. https://www.cdc.gov/nchs/nvss/vsrr/drug-overdose-data.htm . Accessed 2 June 2022.

Multiple Cause of Death 1999–2019 on CDC WONDER Online Database. Centers for disease control and prevention, national center for health statistics. 2020 https://www.cdc.gov/nchs/nvss/vsrr/drug-overdose-data.htm . Accessed 18 May 2021.

Abraham HD, Fava M. Order of onset of substance abuse and depression in a sample of depressed outpatients. Compr Psychiatry. 1999;40(1):44–50. https://doi.org/10.1016/s0010-440x(99)90076-7 .

Article   CAS   PubMed   Google Scholar  

Kessler RC. The epidemiology of dual diagnosis. Biol Psychiatry. 2004;56(10):730–7. https://doi.org/10.1016/j.biopsych.2004.06.034 .

Article   PubMed   Google Scholar  

Merikangas KR, McClair VL. Epidemiology of substance use disorders. Hum Genet. 2012;131(6):779–89. https://doi.org/10.1007/s00439-012-1168-0 .

Wilens TE, Martelon M, Joshi G, Bateman C, Fried R, Petty C, et al. Does ADHD predict substance-use disorders? A 10-year follow-up study of young adults with ADHD. J Am Acad Child Adolesc Psychiatry. 2011;50(6):543–53. https://doi.org/10.1016/j.jaac.2011.01.021 .

Groenman AP, Janssen TWP, Oosterlaan J. Childhood psychiatric disorders as risk factor for subsequent substance abuse: a meta-analysis. J Am Acad Child Adolesc Psychiatry. 2017;56(7):556–69. https://doi.org/10.1016/j.jaac.2017.05.004 .

Russell BS, Trudeau JJ, Leland AJ. Social influence on adolescent polysubstance use: the escalation to opioid use. Subst Use Misuse. 2015;50(10):1325–31. https://doi.org/10.3109/10826084.2015.1013128 .

Shane PA, Jasiukaitis P, Green RS. Treatment outcomes among adolescents with substance abuse problems: The relationship between comorbidities and post-treatment substance involvement. Eval Program Plann. 2003;26(4):393–402. https://doi.org/10.1016/S0149-7189(03)00055-7 .

Article   Google Scholar  

Baker KD, Lubman DI, Cosgrave EM, Killackey EJ, Yuen HP, Hides L, et al. Impact of co-occurring substance use on 6 month outcomes for young people seeking mental health treatment. Aust N Z J Psychiatry. 2007;41(11):896–902. https://doi.org/10.1080/00048670701634986 .

Tolliver BK, Anton RF. Assessment and treatment of mood disorders in the context of substance abuse. Dialogues Clin Neurosci. 2015;17(2):181–90. https://doi.org/10.31887/DCNS.2015.17.2/btolliver .

Mitchell JD, Brown ES, Rush AJ. Comorbid disorders in patients with bipolar disorder and concomitant substance dependence. J Affect Disord. 2007;102(1–3):281–7. https://doi.org/10.1016/j.jad.2007.01.005 .

Curran GM, Sullivan G, Williams K, Han X, Allee E, Kotrla KJ. The association of psychiatric comorbidity and use of the emergency department among persons with substance use disorders: an observational cohort study. BMC Emerg Med. 2008;8:17. https://doi.org/10.1186/1471-227X-8-17 .

Appleby L, Shaw J, Amos T, McDonnell R, Harris C, McCann K, et al. Suicide within 12 months of contact with mental health services: national clinical survey. BMJ. 1999;318(7193):1235–9. https://doi.org/10.1136/bmj.318.7193.1235 .

Article   CAS   PubMed   PubMed Central   Google Scholar  

Oquendo MA, Currier D, Liu SM, Hasin DS, Grant BF, Blanco C. Increased risk for suicidal behavior in comorbid bipolar disorder and alcohol use disorders: results from the national epidemiologic survey on alcohol and related conditions (NESARC). J Clin Psychiatry. 2010;71(7):902–9. https://doi.org/10.4088/JCP.09m05198gry .

Bohnert AS, Ilgen MA, Ignacio RV, McCarthy JF, Valenstein M, Blow FC. Risk of death from accidental overdose associated with psychiatric and substance use disorders. Am J Psychiatry. 2012;169(1):64–70. https://doi.org/10.1176/appi.ajp.2011.10101476 .

Park TW, Lin LA, Hosanagar A, Kogowski A, Paige K, Bohnert AS. Understanding risk factors for opioid overdose in clinical populations to inform treatment and policy. J Addict Med. 2016;10(6):369–81. https://doi.org/10.1097/ADM.0000000000000245 .

Yule AM, Carrellas NW, Fitzgerald M, McKowen JW, Nargiso JE, Bergman BG, et al. Risk factors for overdose in treatment-seeking youth with substance use disorders. J Clin Psychiatry. 2018. https://doi.org/10.4088/JCP.17m11678 .

Wilton G, Stewart LA. Outcomes of offenders with co-occurring substance use disorders and mental disorders. Psychiatr Serv. 2017;68(7):704–9. https://doi.org/10.1176/appi.ps.201500391 .

Gonzalez G, Rosenheck RA. Outcomes and service use among homeless persons with serious mental illness and substance abuse. Psychiatr Serv. 2002;53(4):437–46. https://doi.org/10.1176/appi.ps.53.4.437 .

Saatcioglu O, Yapici A, Cakmak D. Quality of life, depression and anxiety in alcohol dependence. Drug Alcohol Rev. 2008;27(1):83–90. https://doi.org/10.1080/09595230701711140 .

Kronenberg LM, Slager-Visscher K, Goossens PJJ, van den Brink W, van Achterberg T. Everyday life consequences of substance use in adult patients with a substance use disorder (SUD) and co-occurring attention deficit/hyperactivity disorder (ADHD) or autism spectrum disorder (ASD): a patient’s perspective. BMC Psychiatry. 2014;14(1):264. https://doi.org/10.1186/s12888-014-0264-1 .

Olfson M, Shea S, Feder A, Fuentes M, Nomura Y, Gameroff M, et al. Prevalence of anxiety, depression, and substance use disorders in an urban general medicine practice. Arch Fam Med. 2000;9(9):876. https://doi.org/10.1001/archfami.9.9.876 .

Substance Use and Mental Health Services Administration. 2019 National survey of drug use and health (NSDUH) Releases: U.S. department of health and human services. 2019; https://www.samhsa.gov/data/release/2019-national-survey-drug-use-and-health-nsduh-releases . Accessed 2 Aug. 2022.

National Collaborating Centre for Mental Health (UK). Common mental health disorders: identification and pathways to care. british psychological society (UK). 2011 NICE Clinical Guidelines, No. 123. https://www.ncbi.nlm.nih.gov/books/NBK92266/ . Accessed 2 Aug. 2022.

Substance Use and Mental Health Services Administration. white paper on the evidence supporting screening, brief intervention and referral to treatment (SBIRT). 2011. https://www.samhsa.gov/sites/default/files/sbirtwhitepaper_0.pdf . Accessed 2 Aug. 2022.

McNeely J, Kumar PC, Rieckmann T, Sedlander E, Farkas S, Chollak C, et al. Barriers and facilitators affecting the implementation of substance use screening in primary care clinics: a qualitative study of patients, providers, and staff. Addict Sci Clin Pract. 2018;13(1):8. https://doi.org/10.1186/s13722-018-0110-8 .

McPherson TL, Hersch RK. Brief substance use screening instruments for primary care settings: a review. J Subst Abuse Treat. 2000;18(2):193–202. https://doi.org/10.1016/s0740-5472(99)00028-8 .

Pilowsky DJ, Wu L-T. Screening for alcohol and drug use disorders among adults in primary care: a review. Subst Abuse Rehabil. 2012;3:25. https://doi.org/10.2147/SAR.S30057 .

Rahm AK, Boggs JM, Martin C, Price DW, Beck A, Backer TE, et al. Facilitators and barriers to implementing screening, brief intervention, and referral to treatment (SBIRT) in primary care in integrated health care settings. Subst Abus. 2015;36(3):281–8. https://doi.org/10.1080/08897077.2014.951140 .

Paris M Jr, Hoge MA. Burnout in the mental health workforce: a review. J Behav Health Serv Res. 2010;37(4):519–28. https://doi.org/10.1007/s11414-009-9202-2 .

Woltmann EM, Whitley R, McHugo GJ, Brunette M, Torrey WC, Coots L, et al. The role of staff turnover in the implementation of evidence-based practices in mental health care. Psychiatr Serv. 2008;59(7):732–7. https://doi.org/10.1176/ps.2008.59.7.732 .

Gabel S, Radigan M, Wang R, Sederer LI. Health monitoring and promotion among youths with psychiatric disorders: program development and initial findings. Psychiatr Serv. 2011;62(11):1331–7. https://doi.org/10.1176/ps.62.11.pss6211_1331 .

Karno M, Granholm E, Lin A. Factor structure of the alcohol use disorders identification test (audit) in a mental health clinic sample. J Stud Alcohol. 2000;61(5):751–8. https://doi.org/10.15288/jsa.2000.61.751 .

King PR, Beehler GP, Wade M, Buchholz LJ, Funderburk JS, Lilienthal KR, et al. 2018. Opportunities to improve measurement based care practices in mental health care systems a case example of electronic mental health screening and measurement. Fam Syst Health. https://doi.org/10.1037/fsh0000379

Ramo DE, Bahorik AL, Delucchi KL, Campbell CI, Satre DD. Alcohol and drug use, pain and psychiatric symptoms among adults seeking outpatient psychiatric treatment: latent class patterns and relationship to health status. J Psychoactive Drugs. 2018;50(1):43–53. https://doi.org/10.1080/02791072.2017.1401185 .

Satre D, Wolfe W, Eisendrath S, Weisner C. Computerized screening for alcohol and drug use among adults seeking outpatient psychiatric services. Psychiatr Serv. 2008;59(4):441–4. https://doi.org/10.1176/ps.2008.59.4.441 .

Silverman DC, O’Neill SF, Cleary PD, Barwick C, Joseph R. Recognition of alcohol abuse in psychiatric outpatients and its effect on treatment. Psychiatr Serv. 1992;43(6):644–6. https://doi.org/10.1176/ps.43.6.644 .

Article   CAS   Google Scholar  

Denering LL, Spear SE. Routine use of screening and brief intervention for college students in a university counseling center. J Psychoactive Drugs. 2012;44(4):318–24. https://doi.org/10.1080/02791072.2012.718647 .

Stanhope V, Manuel JI, Jessell L, Halliday TM. Implementing SBIRT for adolescents within community mental health organizations: a mixed methods study. J Subst Abuse Treat. 2018;90:38–46. https://doi.org/10.1016/j.jsat.2018.04.009 .

King WM, Restar A, Operario D. Exploring multiple forms of intimate partner violence in a gender and racially/ethnically diverse sample of transgender adults. J Interpers Violence. 2021;36(19–20):10477–98. https://doi.org/10.1177/0886260519876024 .

Bush K, Kivlahan DR, McDonell MB, Fihn SD, Bradley KA, Project ACQI. The AUDIT alcohol consumption questions (AUDIT-C) an effective brief screening test for problem drinking. Arch Intern Med. 1998. https://doi.org/10.1001/archinte.158.16.1789 .

Knight JR, Sherritt L, Shrier LA, Harris SK, Chang G. Validity of the CRAFFT substance abuse screening test among adolescent clinic patients. Arch Pediatr Adolesc Med. 2002;156(6):607–14. https://doi.org/10.1001/archpedi.156.6.607 .

WHO Assist Working Group. The alcohol, smoking and substance involvement screening test (ASSIST): development, reliability and feasibility. Addiction. 2002;97(9):1183–94. https://doi.org/10.1046/j.1360-0443.2002.00185.x .

Selzer ML, Vinokur A, van Rooijen L. A self-administered short Michigan alcoholism screening test (SMAST). J Stud Alcohol. 1975;36(1):117–26. https://doi.org/10.15288/jsa.1975.36.117 .

McNeely J, Adam A, Rotrosen J, Wakeman SE, Wilens TE, Kannry J, et al. Comparison of methods for alcohol and drug screening in primary care clinics. JAMA Netw Open. 2021;4(5):e2110721. https://doi.org/10.1001/jamanetworkopen.2021.10721 .

Substance Abuse and Mental Health Services Administration. National Mental Health Services Survey (N-MHSS) 2020 data on mental health treatment facilities. 2020. https://www.samhsa.gov/data/report/national-mental-health-services-survey-n-mhss-2020-data-mental-health-treatment-facilities . Accessed 2 Aug. 2022.

Sundström C, Petersén E, Sinadinovic K, Gustafsson P, Berman AH. Identification and management of alcohol use and illicit substance use in outpatient psychiatric clinics in Sweden: a national survey of clinic directors and staff. Addict Sci Clin Pract. 2019;14(1):10. https://doi.org/10.1186/s13722-019-0140-x .

Adam A, Schwartz RP, Wu L-T, Subramaniam G, Laska E, Sharma G, et al. Electronic self-administered screening for substance use in adult primary care patients: feasibility and acceptability of the tobacco, alcohol, prescription medication, and other substance use (myTAPS) screening tool. Addict Sci Clin Pract. 2019;14(1):39. https://doi.org/10.1186/s13722-019-0167-z .

Spencer M, Warner M, Bastian BA, Trinidad JP, Hedegaard H. Drug overdose deaths involving fentanyl, 2011–2016. Natl Vital Stat Rep. 2019;68(3):1–19.

PubMed   Google Scholar  

Chisolm DJ, Gardner W, Julian T, Kelleher KJ. Adolescent satisfaction with computer-assisted behavioural risk screening in primary care. Child Adolesc Ment Health. 2008;13(4):163–8. https://doi.org/10.1111/j.1475-3588.2007.00474.x .

Paperny DM, Aono JY, Lehman RM, Hammar SL, Risser J. Computer-assisted detection and intervention in adolescent high-risk health behaviors. J Pediatr. 1990;116(3):456–62. https://doi.org/10.1016/s0022-3476(05)82844-6 .

Rhodes KV, Lauderdale DS, Stocking CB, Howes DS, Roizen MF, Levinson W. Better health while you wait: a controlled trial of a computer-based intervention for screening and health promotion in the emergency department. Ann Emerg Med. 2001;37(3):284–91. https://doi.org/10.1067/mem.2001.110818 .

Jimenez DE, Bartels SJ, Cardenas V, Alegría M. Stigmatizing attitudes toward mental illness among racial/ethnic older adults in primary care. Int J Geriatr Psychiatry. 2013;28(10):1061–8. https://doi.org/10.1002/gps.3928 .

Small J, Curran GM, Booth B. Barriers and facilitators for alcohol treatment for women: are there more or less for rural women? J Subst Abuse Treat. 2010;39(1):1–13. https://doi.org/10.1016/j.jsat.2010.03.002 .

Meirte J, Hellemans N, Anthonissen M, Denteneer L, Maertens K, Moortgat P, et al. Benefits and disadvantages of electronic patient-reported outcome measures: systematic review. JMIR Perioper Med. 2020;3(1):e15588. https://doi.org/10.2196/15588 .

Sisodia RC, Rodriguez JA, Sequist TD. Digital disparities: lessons learned from a patient reported outcomes program during the COVID-19 pandemic. JAMIA Open. 2021;28(10):2265–8. https://doi.org/10.1093/jamia/ocab138 .

Earnshaw VA, Bogart LM, Menino DD, Kelly JF, Chaudoir SR, Reed NM, et al. Disclosure, stigma, and social support among young people receiving treatment for substance use disorders and their caregivers: a qualitative analysis. Int J Ment Health Addict. 2019;17(6):1535–49. https://doi.org/10.1007/s11469-018-9930-8 .

Kelly JF, Greene MC, Abry A. A US national randomized study to guide how best to reduce stigma when describing drug-related impairment in practice and policy. Addiction. 2021;116(7):1757–67. https://doi.org/10.1111/add.15333 .

Simon KM, Harris SK, Shrier LA, Bukstein OG. Measurement-based care in the treatment of adolescents with substance use disorders. Child Adolesc Psychiatr Clin N Am. 2020;29(4):675–90. https://doi.org/10.1016/j.chc.2020.06.006 .

Fadus MC, Squeglia LM, Valadez EA, Tomko RL, Bryant BE, Gray KM. Adolescent substance use disorder treatment: an update on evidence-based strategies. Curr Psychiatry Rep. 2019;21(10):96. https://doi.org/10.1007/s11920-019-1086-0 .

Van Horn DHA, Goodman J, Lynch KG, Bonn-Miller MO, Thomas T, Del Re AC, et al. The predictive validity of the progress assessment, a clinician administered instrument for use in measurement-based care for substance use disorders. Psychiatry Res. 2020. https://doi.org/10.1016/j.psychres.2020.113282 .

Substance Abuse and Mental Health Services Administration. TIP 42: Substance use disorder treatment for people with co-occurring disorders. 2020; 42. https://store.samhsa.gov/product/tip-42-substance-use-treatment-persons-co-occurring-disorders/PEP20-02-01-004 . Accessed 28 July 2022.

Gagnier JJ, Lai J, Mokkink LB, Terwee CB. COSMIN reporting guideline for studies on measurement properties of patient-reported outcome measures. Qual Life Res. 2021. https://doi.org/10.1007/s11136-021-02822-4 .

Download references

Acknowledgements

We would like to acknowledge Melissa Lydston for her essential role in developing our search terms. We would like to acknowledge Sylvia Lanni for her important role in reviewing and editing our manuscript.

AY and TW received funding for this project from the National Institutes of Health through the NIH HEAL Initiative under award number 4UH3DA050252-0. In addition, AY was supported by research funding from the AACAP-NIDA Career Development Award in Substance Use Research (K12), Award Number 5K12DA000357-17, Boston University Doris Duke Charitable Foundation’s Fund to Retain Clinical Scientists, and a Boston University Clinical and Translational Science Institute voucher. She also has funding for clinical program development from the Jack Satter Foundation. CB receives funding from the Harvard Medical School Zinberg Fellowship in Addiction Psychiatry Research, the Massachusetts General Hospital Louis V. Gerstner Research Scholarship, and the AACAP-NIDA Career Development Award in Substance Use Research (K12), Award Number 3K12DA000357-22S1.

Author information

Authors and affiliations.

Department of Psychiatry, Massachusetts General Hospital, Harvard Medical School, Boston, MA, 02114, USA

Diana Woodward, Timothy E. Wilens, Vinod Rao & Colin Burke

Private Clinical Practice, Rockville, MD, USA

Meyer Glantz

Department of Psychiatry, Boston Medical Center, 850 Harrison Avenue, Boston, MA, 02118, USA

Amy M. Yule

You can also search for this author in PubMed   Google Scholar

Contributions

DW, TW, VR, and AY contributed to the development of the study protocol. DW screened the titles and abstracts of all papers. TW and AY reviewed included papers. VR extracted data from quantitative studies. All authors provided contributions to the writing of the manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Amy M. Yule .

Ethics declarations

Ethics approval and consent to participate.

Not applicable.

Consent for publication

Competing interests.

AY is a consultant to the Gavin House and BayCove Human Services (clinical services), as well as the American Psychiatric Association's Providers Clinical Support System Sub-Award. TW has been a consultant for Neurovance/Otsuka, Ironshore, KemPharm, and Vallon, and he has a licensing agreement with Ironshore for a copyrighted diagnostic questionnaire that he co-owns (Before School Functioning Questionnaire). TW also serves as a clinical consultant to the US National Football League (ERM Associates), US Minor/Major League Baseline, Phoenix House/Gavin Foundation, and Bay Cove Human Services. There are no disclosures to report for the remaining authors.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1.

. Search terms used in the systematic review of substance use screening in outpatient behavioral health settings.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Woodward, D., Wilens, T.E., Glantz, M. et al. A systematic review of substance use screening in outpatient behavioral health settings. Addict Sci Clin Pract 18 , 18 (2023). https://doi.org/10.1186/s13722-023-00376-z

Download citation

Received : 25 August 2022

Accepted : 15 March 2023

Published : 26 March 2023

DOI : https://doi.org/10.1186/s13722-023-00376-z

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Substance use
• Behavioral health

Addiction Science & Clinical Practice

ISSN: 1940-0640

• General enquiries: [email protected]

• Fierce Pharma
• Fierce Biotech
• Fierce Healthcare
• Fierce Life Sciences Events
• Cell & Gene Therapy
• Clinical Data
• Venture Capital
• Diagnostics
• AI and Machine Learning
• Special Reports
• Clinical Development
• Special Report
• Awards Gala
• Fierce Events
• Industry Events
• Whitepapers

The Evolving Alzheimer’s Disease Landscape

Alzheimer’s disease research and drug development is evolving at a rapid pace. Decades of research has led to the approval of the first disease-modifying drugs and new pathological discoveries. The first blood-based biomarker tests will bring equitable access to screening and diagnosis while accelerating clinical research. These major breakthroughs in the last few years alone represent a pivotal time in the Alzheimer’s disease landscape. “This is an unprecedented time for Alzheimer’s research. Significant advances have been made in understanding pathological factors affecting disease, long-awaited disease-modifying treatments have been approved and more convenient diagnostic tools are becoming available - all major steps in changing the course and impact of this once-elusive disease,” stated Dr. David Morgan, Director of the Alzheimer’s Alliance and Professor of Translational Neuroscience at Michigan State University. A new era of Alzheimer’s disease therapies There are now ten treatments approved for Alzheimer’s disease, including two recently-approved disease-modifying drugs – the most recent approved in July 2024 . The disease-modifying treatments are a class of monoclonal antibodies that target the removal of amyloid-beta plaque build-up in those with mild cognitive impairment or early Alzheimer’s disease. Researchers convening at the recent Alzheimer’s Association International Conference acknowledged that the community can build upon the latest advances. “We need new therapies, many more targets, better drugs, more convenient drugs, but it is important that we have changed the landscape by having disease-modifying therapies.” Next wave of Alzheimer’s disease testing - plasma-based biomarker tests One of the major advances in the Alzheimer’s disease landscape is the emerging blood-based biomarkers tests that are expected to improve diagnosis in primary care, reduce wait times for treatment initiation, accelerate clinical trial recruitment and greatly lower the cost of diagnosing Alzheimer’s. Only 10 years ago, Alzheimer’s disease diagnosis relied on symptom observations. Since then, the gold standard has moved to a PET scan or lumbar puncture. These diagnostic options are costly, invasive and inaccessible in many communities. One of the challenges to accessing the new monoclonal antibody treatments is the availability of PET scans to screen for qualified patients. The convenience of blood-based biomarker tests would relieve this bottleneck.  A panel has been assigned by the Alzheimer’s Association to conduct a systematic review of the plasma-based biomarker tests and expects to provide a clinical practice guideline in early to mid 2025. Emerging biological pathways as drug development targets Most of the recent drug development efforts have been focused on mechanisms that aim to remove amyloid plaque build-up or tau tangles, main hallmarks of the disease. Researchers are now turning their attention to the role of inflammation in Alzheimer’s disease development and progression. Most recently, a study of a GLP-1 agonist , part of a class of drugs used for diabetes, weight loss and heart disease, demonstrated neuroprotective effects that may help to protect the brains of people with Alzheimer’s disease. Previous studies have shown how GLP-1 agonists can reduce neuroinflammation. The GLP-1 agonists did not alter amyloid-β and tau biomarkers nor show improvements in cognition, but the anti-inflammatory and neuroprotective effects of GLP-1 agonists warrant further studies and opens new biological pathways in the potential treatment of Alzheimer’s. The study showed the potential of repurposing diabetes drugs in the treatment of Alzheimer’s and exploring drug candidates that target similar biological pathways to diabetes treatments. Identifying a combination of therapies or a multi-factorial Approach To treat a complex disease such as Alzheimer’s will likely require a combination of therapies or a multi-factorial approach. The Alzheimer’s Association has identified modifiable risk factors associated with the development and progression of Alzheimer’s disease including untreated visual loss, LDL cholesterol, hearing loss, depression, Type 2 diabetes, exercise and cognitive stimulation. While these preventative lifestyle measures may delay the onset and progression of Alzheimer’s disease, lifestyle habits alone are unlikely to prevent disease development. Clinicians believe that a multi-factorial approach will be required that includes lifestyle habits and a combination of disease-modifying and symptomatic treatments that target multiple biological pathways. InMed’s INM-901 demonstrates a novel, multi-factorial approach to treating Alzheimer’s Among the drug candidates in preclinical development is InMed’s INM-901, a promising small molecule that has demonstrated disease-modifying effects and appears to target multiple biological pathways , including the peroxisome proliferator-activated receptor (PPAR), which is associated with diabetes development and plays an important role in cell metabolism and immune response. Preclinical studies of INM-901 in well-characterized Alzheimer’s disease models indicate that INM-901 reduces neuroinflammation and cell death and promotes the growth of neurites in the brain – which enables cell-to-cell communication and is essential for brain processing. The growth of neurites, which is diminished in patients with Alzheimer’s, signifies enhanced neuronal function and may indicate the potential to restore the damage caused by Alzheimer’s disease. InMed’s INM-901 is among the several promising treatments in development that may be able to change the course of Alzheimer’s disease.

Cholinesterase inhibitors may slow cognitive decline in dementia with Lewy bodies

• Download PDF Copy

Dementia with Lewy bodies is a type of dementia that is similar to both Alzheimer's disease and Parkinson's disease but studies on long-term treatments are lacking. A new study from Karolinska Institutet in Sweden, published in Alzheimer's & Dementia: The Journal of the Alzheimer's Association , highlights the potential cognitive benefits of cholinesterase inhibitor treatment.

Lewy body disease, which includes dementia with Lewy bodies (DLB) and Parkinson's disease with and without dementia, is the second most common neurodegenerative disorder, following Alzheimer's disease. DLB accounts for approximately 10–15 per cent of dementia cases and is characterised by changes in sleep, behaviour, cognition, movement, and regulation of automatic bodily functions.

There are currently no approved treatments for DLB, so doctors often use drugs for Alzheimer's disease, such as cholinesterase inhibitors and memantine, for symptom relief. However, the effectiveness of these treatments remains uncertain due to inconsistent trial results and limited long-term data." Hong Xu, assistant professor at the Department of Neurobiology, Care Sciences and Society, Karolinska Institutet and first author of the paper

In the current study, researchers have examined the long-term effects of cholinesterase inhibitors (ChEIs) and memantine compared with no treatment for up to ten years in 1,095 patients with DLB. They found that ChEIs may slow down cognitive decline over five years compared to memantine or no treatment. ChEIs were also associated with a reduced risk of death in the first year after diagnosis.

"Our results highlight the potential benefits of ChEIs for patients with DLB and support updating treatment guidelines," says Maria Eriksdotter, professor at the Department of Neurobiology, Care Sciences and Society, Karolinska Institutet and last author of the paper.

Related Stories

• New study links brain cell structures to memory loss in aging and dementia
• Violent dementia patients leave nursing home staffers and residents ‘scared to death’
• Study reveals £4bn savings potential from dementia risk reduction interventions

Due to the study's observational nature, no conclusions can be drawn about causality. The researchers did not have data on patient lifestyle habits, frailty, blood pressure, and Alzheimer's disease co-pathology, which may have influenced the findings. Another limitation of the study is that it remains challenging to diagnose DLB accurately.

The research was mainly financed by StratNeuro, the Center for Innovative Medicine (CIMED), the KI foundations and the Swedish Research Council.

Karolinska Institutet

Xu, H.,  et al.  (2024) Long-term effects of cholinesterase inhibitors and memantine on cognitive decline, cardiovascular events, and mortality in dementia with Lewy bodies: An up to 10-year follow-up study . Alzheimer's & Dementia: The Journal of the Alzheimer's Association . doi.org/10.1002/alz.14118 .

Posted in: Medical Research News | Medical Condition News | Pharmaceutical News

Tags: Alzheimer's Disease , Blood , Blood Pressure , Cholinesterase Inhibitor , Dementia , Drugs , Lewy Bodies , Medicine , Mortality , Neurodegenerative Disorder , Parkinson's Disease , Pathology , Research , Sleep

Suggested Reading

Cancel reply to comment

• Trending Stories
• Latest Interviews
• Top Health Articles

Global and Local Efforts to Take Action Against Hepatitis

Lindsey Hiebert and James Amugsi

In this interview, we explore global and local efforts to combat viral hepatitis with Lindsey Hiebert, Deputy Director of the Coalition for Global Hepatitis Elimination (CGHE), and James Amugsi, a Mandela Washington Fellow and Physician Assistant at Sandema Hospital in Ghana. Together, they provide valuable insights into the challenges, successes, and the importance of partnerships in the fight against hepatitis.

Addressing Important Cardiac Biology Questions with Shotgun Top-Down Proteomics

In this interview conducted at Pittcon 2024, we spoke to Professor John Yates about capturing cardiomyocyte cell-to-cell heterogeneity via shotgun top-down proteomics.

A Discussion with Hologic’s Tim Simpson on the Future of Cervical Cancer Screening

Tim Simpson

Hologic’s Tim Simpson Discusses the Future of Cervical Cancer Screening.

Latest News

Newsletters you may be interested in

Your AI Powered Scientific Assistant

Hi, I'm Azthena, you can trust me to find commercial scientific answers from News-Medical.net.

A few things you need to know before we start. Please read and accept to continue.

• Use of “Azthena” is subject to the terms and conditions of use as set out by OpenAI .
• Content provided on any AZoNetwork sites are subject to the site Terms & Conditions and Privacy Policy .
• Large Language Models can make mistakes. Consider checking important information.

Great. Ask your question.

Azthena may occasionally provide inaccurate responses. Read the full terms .

While we only use edited and approved content for Azthena answers, it may on occasions provide incorrect responses. Please confirm any data provided with the related suppliers or authors. We do not provide medical advice, if you search for medical information you must always consult a medical professional before acting on any information provided.

Your questions, but not your email details will be shared with OpenAI and retained for 30 days in accordance with their privacy principles.

Please do not ask questions that use sensitive or confidential information.

Read the full Terms & Conditions .

Provide Feedback

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List

Drug Design and Discovery: Principles and Applications

Shu-feng zhou.

1 Department of Bioengineering and Biotechnology, College of Chemical Engineering, Huaqiao University, Xiamen 361021, Fujian, China

Wei-Zhu Zhong

2 Gordon Life Science Institute, Belmont, MA 02478, USA

Drug discovery is the process through which potential new therapeutic entities are identified, using a combination of computational, experimental, translational, and clinical models (see, e.g., [ 1 , 2 ]). Despite advances in biotechnology and understanding of biological systems, drug discovery is still a lengthy, costly, difficult, and inefficient process with a high attrition rate of new therapeutic discovery. Drug design is the inventive process of finding new medications based on the knowledge of a biological target. In the most basic sense, drug design involves the design of molecules that are complementary in shape and charge to the molecular target with which they interact and bind. Drug design frequently but not necessarily relies on computer modeling techniques and bioinformatics approaches in the big data era. In addition to small molecules, biopharmaceuticals and especially therapeutic antibodies are an increasingly important class of drugs and computational methods for improving the affinity, selectivity, and stability of these protein-based therapeutics have also gained great advances [ 3 ]. Drug development and discovery includes preclinical research on cell-based and animal models and clinical trials on humans, and finally move forward to the step of obtaining regulatory approval in order to market the drug. Modern drug discovery involves the identification of screening hits, medicinal chemistry and optimization of those hits to increase the affinity, selectivity (to reduce the potential of side effects), efficacy/potency, metabolic stability (to increase the half-life), and oral bioavailability. Once a compound that fulfills all of these requirements has been identified, it will begin the process of drug development prior to clinical trials.

This Special Issue “Drug Design and Discovery: Principles and Applications” was focused on the basic principles of modern drug design and discovery and the potential applications. It covered seventeen research articles and one communication contributed from experts all around the world, as briefed below.

The 2014 Ebola epidemic in West Africa is believed to have caused more than 11,000 fatalities. The request for novel drug development, finding efficient drug discovery pathways is going to be crucial in the fight against future outbreaks. In the article entitled “Combating Ebola with Repurposed Therapeutic Using the CANDO Platform” [ 4 ], Gaurav Chopra, Ram Samudrala, and coauthors have developed a Computational Analysis of Novel Drug Opportunities (CANDO) platform based on the hypothesis that drugs function by interacting with multiple protein targets to create a molecular interaction signature that can be exploited for rapid therapeutic repurposing and discovery. They used the CANDO platform to generate top ranking drug candidates for Ebola virus disease treatment, which were compared to those identified from in vitro studies. They found that integrating computational docking predictions on a proteomic scale with results from in vitro screening studies may be used to select and prioritize compounds for further in vivo and clinical testing. This approach will significantly reduce the lead time, risk, cost, and resources required to determine efficacious therapies against future Ebola virus disease outbreaks.

Wei Xiao, Huiming Hua, Jinyi Xu, and their coworkers wrote an article with the title “NO-Releasing Enmein-Type Diterpenoid Derivatives with Selective Antiproliferative Activity and Effects on Apoptosis-Related Proteins” [ 5 ]. They designed and synthesized a series of nine enmein-type ent-kaurane diterpenoid and furoxan-based nitric oxide (NO) donor hybrids from commercially available oridonin. Their investigation in antiproliferative activity of these hybrids suggested that these kinds of NO-donor/diterpenoid hybrids could provide a promising approach for the discovery of novel antitumor agents.

Dimitra Hadjipavlou-Litina and colleagues presented an exhaustive docking analysis considering the case of autotaxin in the article entitled “Boronic Acid Group: A Cumbersome False Negative Case in the Process of Drug Design” [ 6 ]. They found that virtual screening of large libraries of boronic acid derivatives fail to dock in a natural mode. They are left out as false negatives both in regards to their binding poses and their scoring function values. To solve the problems encountered, the authors characterized the formed bond between Ser/Thr residues more accurately as a polar covalent bond instead of as a simple nonpolar covalent bond based on natural bond orbital calculations. The findings described in this article highlight general options that need to be considered when large libraries of boron compounds are virtually screened to identify novel hits in drug design.

In their article “Antiproliferative Activity and Cellular Uptake of Evodiamine and Rutaecarpine Based on 3D Tumor Models” [ 7 ], Feng Xu and coauthors employed the 3D culture of MCF-7 and SMMC-7721 cells based on the hanging drop method and evaluated the anti-proliferative activity and cellular uptake of two promising anti-tumor drug candidates, evodiamine (EVO) and rutaecarpine (RUT), in 3D multicellular spheroids and compared the results with those obtained from 2D monolayers. They believe that their study provided a new vision of the anti-tumor activity of EVO and RUT via 3D multicellular spheroids and cellular uptake through the fluorescence of compounds and may be helpful for drug screening and cytotoxicity studies.

Malaria is one of the principal diseases of developing countries, particularly in Africa, Asia, and South America. Due to the toxic side effects and the risk of developing resistance after prolonged treatment with aminoquinolines, it demands a continuous effort to develop new antimalarial agents, especially as an effective vaccine for malaria is not available. Rizk E. Khidre and colleagues designed and synthesized a novel series of quinoline compounds and screened for their antimalarial activities, with the hope that these compounds could lead to the availability of better drugs to treat malaria. Their study results are presented in the article “New Potential Antimalarial Agents: Design, Synthesis and Biological Evaluation of Some Novel Quinoline Derivatives as Antimalarial Agents” [ 8 ].

As described in the article entitled “Novel ( E )-β-Farnesene Analogues Containing 2-Nitroiminohexahydro-1,3,5-triazine: Synthesis and Biological Activity Evaluation” [ 9 ], Xinling Yang and coauthors introduced a series of novel ( E )-β-farnesene analogues by replacing the conjugated double bonds of EβF with 2-nitroiminohexahydro-1,3,5-triazine. Their bioassay results showed that all the analogues displayed different repellent and aphicidal activities against the green peach aphid (Myzus persicae). They also performed a preliminary structure-activity relationship (SAR), which offered valuable clues for the design of new EβF analogues.

In the search for prodrug analogs of clopidogrel with improved metabolic characteristics and antiplatelet bioactivity, a group of clopidogrel and vicagrel analogs selectively deuterated at the benzylic methyl ester group were synthesized, characterized, and evaluated by Yan Yang, Jingkai Gu, and their colleagues. The ability of the compounds to inhibit ADP-induced platelet aggregation and pharmacokinetics from rats after oral dosing were studied and the results are detailed in the article “Significant Improvement of Metabolic Characteristics and Bioactivities of Clopidogrel and Analogs by Selective Deuteration” [ 10 ].

Interest in intranasal administration as a non-invasive route for drug delivery continues to grow rapidly. Because of the sensitive respiratory mucosa, not only the active ingredients, but also additives need to be tested in appropriate models for toxicity. Rita Ambrus and coworkers studied the cytotoxicity of six pharmaceutical excipients, which could help to reach larger residence time, better permeability, and an increased solubility dissolution rate. As described in the communication entitled “Cytotoxicity of Different Excipients on RPMI 2650 Human Nasal Epithelial Cells” [ 11 ], they found that all additives at 0.3% sodium hyaluronate and polyvinyl alcohol at 1% concentrations can be safely used for nasal formulations.

As spermatozoa become mature and acquire fertilizing capacity during their passage through the epididymal lumen, Li-Juan Qu, Yan Zhu, et al. conducted a study to identify new epididymal luminal fluid proteins involved in sperm maturation in infertile rats by dutasteride, a dual 5α-reductase inhibitor, in order to provide potential epididymal targets for new contraceptives and infertility treatments. They report for the first time that dutasteride influences the protein expression profiling in the epididymal luminal fluids of rats, and this result provides some new epididymal targets for male contraception and infertility therapy. The study results are presented in the article “Identification of New Epididymal Luminal Fluid Proteins Involved in Sperm Maturation in Infertile Rats Treated by Dutasteride Using iTRAQ” [ 12 ].

Reported in the article “Synthesis and Evaluation of Ester Derivatives of 10-Hydroxycanthin-6-one as Potential Antimicrobial Agents” [ 13 ], Jun-Ru Wang and coauthors studied a new series of ester derivatives of 10-hydroxycanthin-6-one using a simple and effective synthetic route as part of their continuing research on canthin-6-one antimicrobial agents. They characterized the structure and antimicrobial activity of each compound, investigated the structure-activity relationship, and identified the promising lead compound that had significant antimicrobial activity against all the fungi and bacterial strains tested for the development of novel canthine-6-one antimicrobial agents.

Chun-Mei Jin, Zhe-Shan Quan, and colleagues wrote an article entitled “Synthesis and Biological Evaluation of Novel Benzothiazole Derivatives as Potential Anticonvulsant Agents” [ 14 ]. Because of the crucial need to develop more effective antiepileptic drugs endowed with an improved safety profile, the authors investigated new benzotriazoles with a mercapto-triazole and other heterocycle substituents, and evaluated their anticonvulsant activity and neurotoxicity for each compound by using the maximal electroshock, subcutaneous pentylenetetrazole, and rotarod neurotoxicity tests. The study outcomes are presented in their paper [ 14 ].

Non-steroidal anti-inflammatory drugs are the most commonly prescribed anti-inflammatory and pain relief medications. However, their use is associated with many drawbacks. In an attempt to circumvent these risks, Ahmed M. Gouda and coworkers designed, synthesized, and evaluated a set of N-(4-bromophenyl)-7-cyano-6-substituted-H-pyrrolizine-5-carboxamide derivatives as dual COX/5-LOX inhibitors. In light of their findings, these novel pyrrolizine-5-carboxamide derivatives represent a promising scaffold for further development into potential dual COX/5-LOX inhibitors with safer gastric profiles. Their results are detailed in the article “Design, Synthesis, and Biological Evaluation of Some Novel Pyrrolizine Derivatives as COX Inhibitors with Anti Inflammatory/Analgesic Activities and Low Ulcerogenic Liability” [ 15 ].

The main step in a successful drug discovery pipeline is the identification of small potent compounds that selectively bind to the target of interest with high affinity. In the work reported in the article “Self Organizing Map-Based Classification of Cathepsin k and S Inhibitors with Different Selectivity Profiles Using Different Structural Molecular Fingerprints: Design and Application for Discovery of Novel Hits” [ 16 ], Hany E. A. Ahmed and colleagues proposed an affordable machine learning method to perform compound selectivity classification and prediction. They collected compounds with reported activity and built a selectivity database formed of 153 cathepsin K and S inhibitors that are considered of medicinal interest. The study results exhibited the capability of the method in the design of further novel inhibitors with high activity and target selectivity.

Vancomycin, a widely used antibiotic, often induces nephrotoxicity; however, the molecular targets and underlying mechanisms of this side effect remain unclear. In order to uncover the comprehensive and global understanding on the effect of vancomycin, Zhi-Ling Li and Shu-Feng Zhou investigated the molecular targets of vancomycin in human proximal tubule epithelial HK-2 cells with a focus on cell cycle, apoptosis, autophagy, and epithelial to mesenchymal transition (EMT) pathways. The quantitative SILAC-based proteomic approach showed that vancomycin regulated cell proliferation, mitochondria-dependent apoptotic pathway and autophagy, and EMT in HK-2 cells, involving a number of key functional proteins and related molecular signaling pathways. This study may provide a clue to fully identify the molecular targets and elucidate the underlying mechanism of vancomycin-associated nephrotoxicity, resulting in an improved therapeutic effect and reduced side effect in clinical settings. Detailed results are presented in the article “A SILAC-Based Approach Elicits the Proteomic Responses to Vancomycin-Associated Nephrotoxicity in Human Proximal Tubule Epithelial HK-2 Cells” [ 17 ].

Knowledge of protein-protein interactions and their binding sites is indispensable for in-depth understanding of the networks in living cells. With the avalanche of protein sequences generated in the postgenomic age, it is critical to develop computational methods for identifying in a timely fashion the protein-protein binding sites (PPBSs) based on the sequence information alone because the information obtained by this method can be used for both biomedical research and drug development. To address such a challenge, Jianhua Jia, Bingxiang Liu, and colleagues [ 18 ] have proposed a new predictor, called iPPBS-Opt, in which they have used the concept of pseudo amino acid composition (PseAAC) [ 19 ] to formulate complicated protein sequences. Although there are many investigators (see, e.g., [ 20 , 21 , 22 , 23 ]) who also used the PseAAC to formulate protein sequences, this is the first time the stationary wavelet transform approach is introduced to reflect the functions of low-frequency phonons in proteins as deduced some 40 years ago [ 24 , 25 ]. Furthermore, to maximize the convenience for most experimental scientists, they have provided a step-by-step guide on how to use the predictor’s web server ( http://www.jci-bioinfo.cn/iPPBS-Opt ) to obtain the desired results without the need to go through the complicated mathematical equations involved.

In the article “Synthesis of Canthardin Sulfanilamides and Their Acid Anhydride Analogues via a Ring-Opening Reaction of Activated Aziridines and Their Associated Pharmacological Effects” [ 26 ], Mei-Hsiang Lin and coworkers reported their investigation to find new cantharidinimides and related imides containing the sulfonamide group. The modification of cantharidinimide by means of the reaction of activated aziridine ring opening led to the discovery of a novel class of antitumor compounds. They found that the most potent cytostatic compound, N -cantharidinimido-sulfamethazine, exhibited anti-HL-60 and anti-Hep3B cell activities. Detailed results of their investigation are presented in the article [ 26 ].

Jian Li and coworkers wrote an article entitled “Chemical Structure-Related Drug-Like Criteria of Global Approved Drugs” [ 27 ]. They uncovered three important structure-related criteria closely related to drug-likeness, namely: (1) the best numbers of aromatic and non-aromatic rings are 2 and 1, respectively; (2) the best functional groups of candidate drugs are usually -OH, -COOR, and -COOH in turn, but not -CONHOH, -SH, -CHO, and -SO 3 H. In addition, the -F functional group is beneficial to CNS drugs, and the -NH 2 functional group is beneficial to anti-infective drugs and anti-cancer drugs; (3) the best R value intervals of candidate drugs are in the range of 0.05–0.50 (preferably 0.10–0.35), and the R value of candidate CNS drugs should be as small as possible in this interval. They envision that the three chemical structure-related criteria may be applicable in a prospective manner for the identification of novel candidate drugs and will provide a theoretical foundation for designing new chemical entities with good drug-like properties.

For the purpose of finding highly active pyrazole amide compounds, Jin-Xia Mu, Xing-Hai Liu, Bao-Ju Li, and their coworkers designed and synthesized a series of novel pyrazole amide derivatives by multi-step reactions from phenylhydrazine and ethyl 3-oxobutanoate as starting materials. They characterized the structures and antifungal activities of the title compounds and used DFT calculations to study the structure-activity relationships. Their results indicated that some of the title compounds exhibited moderate antifungal activity, as shown in the article “Design, Synthesis, DFT Study and Antifungal Activity of Pyrazolecarboxamide Derivatives” [ 28 ].

The eighteen articles published in this thematic issue “Drug Design and Discovery: Principles and Applications” are highlighted in the areas of computer-aided drug discovery and development, drug design and synthesis approaches, in vitro and in vivo pharmacological and toxicological evaluations, etc. These articles not only provided important information, but also generated many useful tools for drug discovery and development. These works showed that the in vitro and in vivo experiments complemented with computation methods are continuously improving the effectiveness and efficiency of drug discovery processes to select lead candidates with more favorable pharmacological, pharmacokinetics, and toxicological profiles.

It is our intent that publication of this Special Issue can stimulate new strategies in drug design and provide new tools, approaches, and technologies to facilitate the evaluation of new drug candidates, leading to the rapid and successful development of novel, effective, and safe medicines for treating diseases [ 29 ].

Conflicts of Interest

The authors declare no conflict of interest.