Graph-convolution neural network-based flexible docking utilizing coarse-grained distance matrix

TL;DR

This paper introduces two innovative deep neural network methods leveraging graph convolutional networks to improve the efficiency and accuracy of predicting flexible protein-ligand binding modes, implicitly accounting for side-chain flexibility.

Contribution

It presents novel neural network approaches that outperform standard docking methods in predicting flexible protein-ligand complexes, incorporating side-chain flexibility via distance matrix prediction.

Findings

Significant improvement in binding mode prediction accuracy.

Enhanced efficiency over standard docking methods.

Effective handling of side-chain flexibility through distance matrix prediction.

Abstract

Prediction of protein-ligand complexes for flexible proteins remains still a challenging problem in computational structural biology and drug design. Here we present two novel deep neural network approaches with significant improvement in efficiency and accuracy of binding mode prediction on a large and diverse set of protein systems compared to standard docking. Whereas the first graph convolutional network is used for re-ranking poses the second approach aims to generate and rank poses independent of standard docking approaches. This novel approach relies on the prediction of distance matrices between ligand atoms and protein C_alpha atoms thus incorporating side-chain flexibility implicitly.