DeepDist: real-value inter-residue distance prediction with deep residual convolutional network

Abstract

Background Driven by deep learning, inter-residue contact/distance prediction has been significantly improved and substantially enhanced ab initio protein structure prediction. Currently, most of the distance prediction methods classify inter-residue distances into multiple distance intervals instead of directly predicting real-value distances. The output of the former has to be converted into real-value distances to be used in tertiary structure prediction. Results To explore the potentials of predicting real-value inter-residue distances, we develop a multi-task deep learning distance predictor (DeepDist) based on new residual convolutional network architectures to simultaneously predict real-value inter-residue distances and classify them into multiple distance intervals. Tested on 43 CASP13 hard domains, DeepDist achieves comparable performance in real-value distance prediction and multi-class distance prediction. The average mean square error (MSE) of DeepDist’s real-value distance prediction is 0.896 Å2 when filtering out the predicted distance ≥ 16 Å, which is lower than 1.003 Å2 of DeepDist’s multi-class distance prediction. When distance predictions are converted into contact predictions at 8 Å threshold (the standard threshold in the field), the precision of top L/5 and L/2 contact predictions of DeepDist’s multi-class distance prediction is 79.3% and 66.1%, respectively, higher than 78.6% and 64.5% of its real-value distance prediction and the best results in the CASP13 experiment. Conclusions DeepDist can predict inter-residue distances well and improve binary contact prediction over the existing state-of-the-art methods. Moreover, the predicted real-value distances can be directly used to reconstruct protein tertiary structures better than multi-class distance predictions due to the lower MSE. Finally, we demonstrate that predicting the real-value distance map and multi-class distance map at the same time performs better than predicting real-value distances alone.

DISTEMA: distance map-based estimation of single protein model accuracy with attentive 2D convolutional neural network

“…We applied a real-value distance predictor DeepDist [18] to predict an inter-residue distance map from the sequence of a protein target as matrix A ( L × L ), where L donates the sequence length and A [ i,j ] is the distance between residues i and j . A was compared with the distance matrix B ( L × L ) calculated from the coordinates of residues in a protein structure model to generate a difference map D .…”

Recent Advances in Protein Homology Detection Propelled by Inter-Residue Interaction Map Threading

“…Recent studies ( Xu and Wang, 2019 ; Xu, 2019 ) have demonstrated the advantage of using distance maps in protein structure prediction over binary contacts as distances carry more physical constraint information of protein structures than contacts. The granularities of predicted distance maps vary from distance histograms to real-valued distances ( Greener et al, 2019 ; Adhikari, 2020 ; Ding and Gong, 2020 ; Li and Xu, 2020 ; Wu et al, 2021 ; Yang et al, 2020 ). Very recently, trRosetta ( Yang et al, 2020 ) has introduced inter-residue orientations in addition to distances to capture not only the spatial proximity information of the interacting pairs but also their relative angles and dihedrals.…”

A deep dilated convolutional residual network for predicting interchain contacts of protein homodimers

“…The computational methods for predicting protein tertiary structures and quaternary structures are periodically evaluated in the Critical Assessment of Protein Structure Prediction (CASP) (Kryshtafovych et al, 2014;Moult et al, 2016;Kryshtafovych et al, 2019;Kwon et al, 2021) and the Critical Assessment of Protein Interaction (CAPRI) (Lensink et al, 2016(Lensink et al, , 2018(Lensink et al, , 2021, respectively, or the joint experiment of the two. Driven by the application of deep learning methods to predicting residueresidue contacts and distances (Wang et al, 2017;Adhikari et al, 2018;Jones & Kandathil, 2018;Li et al, 2019;Hou et al, 2020;Senior et al, 2020;Yang et al, 2020;Wu et al, 2021) in the last several years, tertiary structure prediction has reached unprecedented high accuracy. In the 2020 CASP14 experiment, AlphaFold2 (Jumper et al, 2021) predicted high-quality structures for most CASP14 targets with the accuracy equal to or close to that of the experimental structure determination.…”