Tree-Based Methods to Predict Enzyme Inhibition

Machine learning techniques contribute to studying complex systems by building robust models to predict their behavior. The intrinsic complexity of biological systems (including proteins) makes them viable targets for machine learning modeling. This chapter describes computational models constructed to predict the inhibition of cyclin-dependent kinase 4 (CDK4). This enzyme is a cell-cycle regulator and a target for anticancer drugs. Application of tree-based methods (Decision Trees, Extra Trees, and Random Forest) to build regression models focused on CDK4 produced a model with superior predictive performance. An Extra Trees model outperforms a classical scoring function available in docking programs (e.g., Plants score). The building of the regression models employed features determined using the program Molegro Virtual Docker (MVD). The Jupyter Notebook SKReg4Model built all the models discussed here. This code has 80 regression methods implemented using the Scikit-Learn library. All CDK4 datasets and Jupyter Notebooks discussed in this work are available at GitHub: https://github.com/azevedolab/docking#readme .