Predicting protein variants with equivariant graph neural networks

TL;DR

This study compares structure-based equivariant graph neural networks and sequence-based models for predicting beneficial protein variants, finding that structural methods perform competitively with less data and that combining assay data with pre-trained models enhances prediction.

Contribution

It provides the first comparative analysis of structure- and sequence-based methods for protein variant prediction using EGNNs and demonstrates the effectiveness of structural approaches with limited data.

Findings

Structural approach achieves competitive performance with fewer molecules.

Combining assay data with pre-trained models improves predictions.

Structural methods are viable alternatives to sequence-based models.

Abstract

Pre-trained models have been successful in many protein engineering tasks. Most notably, sequence-based models have achieved state-of-the-art performance on protein fitness prediction while structure-based models have been used experimentally to develop proteins with enhanced functions. However, there is a research gap in comparing structure- and sequence-based methods for predicting protein variants that are better than the wildtype protein. This paper aims to address this gap by conducting a comparative study between the abilities of equivariant graph neural networks (EGNNs) and sequence-based approaches to identify promising amino-acid mutations. The results show that our proposed structural approach achieves a competitive performance to sequence-based methods while being trained on significantly fewer molecules. Additionally, we find that combining assay labelled data with structure pre-trained models yields similar trends as with sequence pre-trained models. Our code and trained models can be found at: https://github.com/semiluna/partIII-amino-acid-prediction.