Boosting Convolutional Neural Networks' Protein Binding Site Prediction Capacity Using SE(3)-invariant transformers, Transfer Learning and Homology-based Augmentation

TL;DR

This paper introduces a novel deep learning approach for predicting protein binding sites, leveraging SE(3)-invariant transformers, transfer learning, and homology-based augmentation, achieving state-of-the-art results in both pocket and residue resolution predictions.

Contribution

It presents a new architecture with SE(3)-invariant self-attention, transfer learning between resolutions, and a homology-based augmentation method, advancing protein binding site prediction techniques.

Findings

Outperformed all state-of-the-art baselines in binding site prediction

Demonstrated good performance on a case study with human serum albumin

Model effectively used transfer learning and homology augmentation to improve accuracy

Abstract

Figuring out small molecule binding sites in target proteins, in the resolution of either pocket or residue, is critical in many virtual and real drug-discovery scenarios. Since it is not always easy to find such binding sites based on domain knowledge or traditional methods, different deep learning methods that predict binding sites out of protein structures have been developed in recent years. Here we present a new such deep learning algorithm, that significantly outperformed all state-of-the-art baselines in terms of the both resolutions$\unicode{x2013}$pocket and residue. This good performance was also demonstrated in a case study involving the protein human serum albumin and its binding sites. Our algorithm included new ideas both in the model architecture and in the training method. For the model architecture, it incorporated SE(3)-invariant geometric self-attention layers that operate on top of residue-level CNN outputs. This residue-level processing of the model allowed a transfer learning between the two resolutions, which turned out to significantly improve the binding pocket prediction. Moreover, we developed novel augmentation method based on protein homology, which prevented our model from over-fitting. Overall, we believe that our contribution to the literature is twofold. First, we provided a new computational method for binding site prediction that is relevant to real-world applications, as shown by the good performance on different benchmarks and case study. Second, the novel ideas in our method$\unicode{x2013}$the model architecture, transfer learning and the homology augmentation$\unicode{x2013}$would serve as useful components in future works.