QDeep: distance-based protein model quality estimation by residue-level ensemble error classifications using stacked deep residual neural networks

Abstract

Motivation:Protein model quality estimation, in many ways, informs protein structure prediction. Despite their tight coupling, existing model quality estimation methods do not leverage inter-residue distance information or the latest technological breakthrough in deep learning that has recently revolutionized protein structure prediction. Results: We present a new distance-based single-model quality estimation method called QDeep by harnessing the power of stacked deep residual neural networks (ResNets). Our method first employs stacked deep ResNets to perform residue-level ensemble error classifications at multiple predefined error thresholds, and then combines the predictions from the individual error classifiers for estimating the quality of a protein structural model. Experimental results show that our method consistently outperforms existing state-of-the-art methods including ProQ2, ProQ3, ProQ3D, ProQ4, 3DCNN, MESHI, and VoroMQA in multiple independent test datasets across a wide-range of accuracy measures; and that predicted distance information significantly contributes to the improved performance of QDeep.

Improved protein model quality assessment by integrating sequential and pairwise features using deep learning

“…We compare our method with three leading single-model QA methods: ProQ4 (Hurtado et al, 2018), CNNQA (Hou et al, 2019) and QDeep (Shuvo et al, 2020). Meanwhile, ProQ4 performed very well in CASP13 and QDeep predicts only global quality.…”

“…CNNQA applies 1D CNN to predict local and global model quality from sequential features, Rosetta energy terms, and knowledge-based potentials (Hou et al, 2019). QDeep (Shuvo et al, 2020) uses an ensemble of stacked 1D CNNs to predict quality from predicted distance information and some similar sequential features used by CNNQA. These methods mainly use coarse-grained structure representation and thus, may not perform well on local QA.…”

Protein Model Accuracy Estimation Empowered by Deep Learning and Inter-residue Distance Prediction in CASP14

“…In the 13th Critical Assessment of Protein Structure Prediction (CASP13), the inter-residue contact information and deep learning were the key for DeepRank 1 to achieve the best performance in ranking protein structural models with the minimum loss of GDT-TS score 2 . Recently, inter-residue distance predictions have been used with deep learning for the estimation of model accuracy 3,4,5 .…”

ROPIUS0: A deep learning-based protocol for protein structure prediction and model selection and its performance in CASP14

“…More generally, given the target sequence and a set of structural models, a model selection algorithm aims to select the most accurate model, and, therefore, it is as important as discriminating between correct and incorrect protein folds. In the context of independent model accuracy estimation (Olechnovič and Venclovas, 2017;Studer et al, 2020), the progress in deep learning has fostered the application of deep NNs to predicting model accuracy locally for each residue (Hurtado et al, 2018;Shuvo et al, 2020).…”

“…CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted June 22, 2021. ; https://doi.org/10.1101/2021.06.22.449457 doi: bioRxiv preprint 2018; Shuvo et al, 2020).…”