Predicting the binding affinities of drug-protein interaction by analyzing the images of binding sites

Thesis Type: Doctorate

Institution Of The Thesis: Orta Doğu Teknik Üniversitesi, Faculty of Engineering, Department of Computer Engineering, Turkey

Approval Date: 2013




Analysis of protein-ligand interactions plays an important role in designing safe and efficient drugs, contributing to drug discovery and development. Recently, machine learning methods have been found useful in drug design, which utilize intelligent techniques to predict unknown protein-ligand interactions by learning from specific properties of known protein-ligand interactions. The aim of this thesis is to propose a novel computational model, Compressed Images for Affinity Prediction (CIFAP), to predict binding affinities of structurally related protein-ligand complexes. The novel method presented here is based on a protein-ligand model from which computational affinity information is obtained by utilizing 2D electrostatic potential images determined for the binding site of the proteins with its inhibitors. The patterns obtained from the 2D images were used for building a predictive model whose strength was tested using Partial Least Squares Regression (PLSR), Support Vector Regression (SVR) and Adaptive Neuro-Fuzzy Inference System (ANFIS) in comparison. The experiments were conducted on two distinct protein-ligand complex systems, which were complexes of CHK1-thienopyridine derivatives and CASP3-isatin sulfonamide derivatives. It is observed that the pixels of the images which are close to the surfaces of the interaction site have better explanation of the binding affinity. Moreover, PLSR is found to be the most promising prediction method for CIFAP as compared to SVR and ANFIS with the lowest error and the highest correlation between the observed and experimental binding affinities. The Computational algorithm presented here is proposed to have a great potential in pharmacophore-based drug design, especially in prediction of binding related properties.