MolSHAP: Interpreting Quantitative Structure–Activity Relationships Using Shapley Values of R-Groups

Abstract

Optimizing the activities and properties of lead compounds is an essential step in the drug discovery process. Despite recent advances in machine learning-aided drug discovery, most of the existing methods focus on making predictions for the desired objectives directly while ignoring the explanations for predictions. Although several techniques can provide interpretations for machine learning-based methods such as feature attribution, there are still gaps between these interpretations and the principles commonly adopted by medicinal chemists when designing and optimizing molecules. Here, we propose an interpretation framework, named MolSHAP, for quantitative structure−activity relationship analysis by estimating the contributions of R-groups. Instead of attributing the activities to individual input features, MolSHAP regards the R-group fragments as the basic units of interpretation, which is in accordance with the fragment-based modifications in molecule optimization. MolSHAP is a model-agnostic method that can interpret activity regression models with arbitrary input formats and model architectures. Based on the evaluations of numerous representative activity regression models on a specially designed R-group ranking task, MolSHAP achieved significantly better interpretation power compared with other methods. In addition, we developed a compound optimization algorithm based on MolSHAP and illustrated the reliability of the optimized compounds using an independent case study. These results demonstrated that MolSHAP can provide a useful tool for accurately interpreting the quantitative structure−activity relationships and rationally optimizing the compound activities in drug discovery.

Generalizability Improvement of Interpretable Symbolic Regression Models for Quantitative Structure–Activity Relationships

“…Interpretation approaches can be classified into two categories: global and local, corresponding to focusing on the prediction model itself and input instances, respectively. Various studies were reported on the local interpretation methods and associated applications, such as the identification of important substructures (atoms) in a molecule, − compound optimization , and finding a potential artifact of a model . On the other hand, methods for global interpretations are limited.…”

Editorial: Machine Learning in Bio-cheminformatics

“…( b ) virtual screening and drug design . Some methods, algorithms, and functional tools were constructed to facilitate the application or improve the performance of the classic virtual screening strategy. − Machine learning methods were also adopted in this collection to identify new hit compounds, discover promising leads for cholestasis, interpret QSAR models, and learn molecular representations. − DiStefano et al and Mao et al conducted research on toxicity prediction and antiviral drug design, respectively. Moreover, ML was also applied to explore the pharmaceutical properties of diverse drug candidates. − A novel knowledge base for nonalcoholic fatty liver disease was developed .…”

Efficient Deep Model Ensemble Framework for Drug-Target Interaction Prediction

“…Recently, Shapley values have been expansively adopted in the realm of machine learning, especially in elucidating model decisions. Within the ambit of substructure importance analysis, conventional methodologies often perceive substructures as autonomous entities, determining their relevance based on model predictions.…”