Protein tertiary structure modeling driven by deep learning and contact distance prediction in CASP13

Abstract

Prediction of residue-residue distance relationships (e.g. contacts) has become the key direction to advance protein tertiary structure prediction since 2014 CASP11 experiment, while deep learning has revolutionized the technology for contact and distance distribution prediction since its debut in 2012 CASP10 experiment. During 2018 CASP13 experiment, we enhanced our MULTICOM protein structure prediction system with three major components: contact distance prediction based on deep convolutional neural networks, contact distance-driven template-free (ab initio) modeling, and protein model ranking empowered by deep learning and contact prediction, in addition to an update of other components such as template library, sequence database, and alignment tools. Our experiment demonstrates that contact distance prediction and deep learning methods are the key reasons that MULTICOM was ranked 3rd out of all 98 predictors in both template-free and template-based protein structure modeling in CASP13. Deep convolutional neural network can utilize global information in pairwise residue-residue features such as co-evolution scores to substantially improve inter-residue contact distance prediction, which played a decisive role in correctly folding some free modeling and hard template-based modeling targets from scratch. Deep learning also successfully integrated 1D structural features, 2D contact information, and 3D structural quality scores to improve protein model quality assessment, where the contact prediction was demonstrated to consistently enhance ranking of protein models for the first time. The success of MULTICOM system in the CASP13 experiment clearly shows that protein contact distance prediction and model selection driven by powerful deep learning holds the key of solving protein structure prediction problem. However, there are still major challenges in accurately predicting protein contact distance when there are few homologous sequences to generate co-evolutionary signals, folding proteins from noisy contact distances, and ranking models of hard targets.