Computational studies of structure and function of viral reverse transcriptase

Effects of amino acid residue mutations on the enzymatic activity and drug resistance of HIV reverse transcriptase (HIV RT) has been studied using computational geometry and machine learning approaches. Residual scores for protein mutants were calculated from the four-body statistical potential based on the Delaunay tessellation of protein structure. One-dimensional profiles of residual scores for experimentally studied HIV RT mutants with known activity and drug resistance were used as a training set for several supervised machine-learning algorithms, including Support Vector Machine (SVM), Decision Tree (DT) and Random Forest (RF). The resulting models can be utilized for screening of all possible mutants for desired activity and drug resistance with high accuracy.