Predicting Drug Sensitivity of Cancer Cell Lines to BET Inhibitor OTX015 Using Machine Learning Approaches

Bromodomain and Extra-Terminal (BET) proteins are crucial epigenetic regulators involved in transcriptional processes linked to cancer progression. Inhibiting these proteins has emerged as a promising therapeutic strategy, with OTX015, a potent BET inhibitor, showing promising efficacy against various cancer types. However, accurately predicting drug sensitivity across cancer cell lines remains a challenge. In this study, we present a machine learning-based approach to predict the half-maximal inhibitory concentration (IC₅₀) of OTX015 across various cancer cell lines using their gene expression profiles. By employing regression-based machine learning models, including Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Extreme Gradient Boosting (XGB), Elastic Net (EN), and Neural Networks (NNET), we aimed to optimize prediction accuracy. The dataset was divided into subsets containing 10, 25, 50, 75, and 100 genes based on their correlation with IC₅₀ values, enabling a comprehensive evaluation of model performance across different data dimensions. The SVM model consistently demonstrated the best performance, achieving the lowest Mean Absolute Error (MAE) scores across all datasets, thereby proving most effective for predicting IC₅₀ values. This approach highlights the potential of integrating machine learning algorithms with gene expression data to enhance drug discovery and personalized medicine, particularly in the context of cancer research. Future work should focus on expanding datasets, optimizing feature selection, and evaluating additional machine learning approaches to improve prediction reliability and generalizability.