Quantitative Structure-Mutation-Activity Relationship Tests (QSMART) Model for Protein Kinase Inhibitor Response Prediction

Predicting how mutations impact drug sensitivity is a major challenge in personalized medicine. Although several machine learning models have been developed to predict drug sensitivity from gene expression and genomic profiles, these methods do not explicitly incorporate the structural properties of drug-mutation interactions to understand the molecular mechanisms of drug resistance/sensitivity. To facilitate the understanding of how the drug-mutation interactions quantitatively contribute to drug response, we developed a framework that not only estimates IC50 with high accuracy (R-squared = 0.861 and RMSE = 0.818) but also identifies features contributing to the accuracy, thereby enhancing explainability. Our framework uses a multi-component approach that includes (1) collecting drug fingerprints, cancer cell line9s multi-omics features, and drug responses, (2) testing the statistical significance of interaction effects, (3) selecting features by Lasso with Bayesian information criterion, and (4) using neural networks to predict drug response. We validate each component in the proposed framework and explain the biological relevance and mathematical interpretation of pertinent features, including afatinib- and lapatinib-EGFR L858R interactions, in a non-small cell lung cancer case study. This is the first study to systematically explain drug response in cancer cell lines by investigating the contribution of interaction effects, such as protein-protein interactions and drug-mutation interactions. The concept of our proposed framework can also be applied to other prediction models with the interaction effects of interest, such as drug-drug interaction and agent-host interaction.