PYTHIA: Deep Learning Approach for Local Protein Conformation Prediction.

Gabriel Cretin,Alexandre G De Brevern,Tatiana Galochkina,Jean-Christophe Gelly

doi:10.3390/ijms22168831

Abstract

Protein Blocks (PBs) are a widely used structural alphabet describing local protein backbone conformation in terms of 16 possible conformational states, adopted by five consecutive amino acids. The representation of complex protein 3D structures as 1D PB sequences was previously successfully applied to protein structure alignment and protein structure prediction. In the current study, we present a new model, PYTHIA (predicting any conformation at high accuracy), for the prediction of the protein local conformations in terms of PBs directly from the amino acid sequence. PYTHIA is based on a deep residual inception-inside-inception neural network with convolutional block attention modules, predicting 1 of 16 PB classes from evolutionary information combined to physicochemical properties of individual amino acids. PYTHIA clearly outperforms the LOCUSTRA reference method for all PB classes and demonstrates great performance for PB prediction on particularly challenging proteins from the CASP14 free modelling category.

Highlights

PYTHIA performs prediction using a deep neural network trained on a non-redundant data set of protein structures
We report a deep learning-based model for protein local conformation prediction in terms of Protein Blocks
PYTHIA demonstrates an important improvement of prediction performance over the reference support vector machine (SVM)-based method LOCUSTRA

Summary

Introduction

Protein structure can be described at different levels of granulometry. Local protein organization at the residue level is described in terms of secondary structures: α-helices and β-strands. The assignment of regular secondary structures is based on the pattern of hydrogen bonds between the amino hydrogen and carboxyl oxygen atoms in the protein backbone and represents nearly fifty percent of protein residues on average. All the unassigned protein regions are classified as coils. While such a description provides essential information on protein structural local conformation, it lacks precision. A more complete secondary structure classification was implemented by the Define Secondary

Methods

Results

Conclusion