Combining pairwise structural similarity and deep learning interface contact prediction to estimate protein complex model accuracy in CASP15

Roy, Raj; Liu, Jian; Giri, Nabin; Guo, Zhiye; Cheng, Jianlin

doi:10.1002/prot.26542

Cited by 9 publications

(10 citation statements)

References 54 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…DeepUMQA3 uses the average of the lDDT of all residues (global lDDT) as the overall fold accuracy, and GraphGPSM directly predicts the TM‐score of the overall complex. The comprehensive performance of GraphGPSM and DeepUMQA3 ranked seventh and ninth among the 23 participating methods, respectively, and MULTICOM_qa 42 ranked first. In this section, TM‐score and global lDDT are used to measure the global accuracy evaluation performance of DeepUMQA3 and GraphGPSM.…”

Section: Resultsmentioning

confidence: 99%

Estimating protein complex model accuracy based on ultrafast shape recognition and deep learning in CASP15

Liu

et al. 2023

Proteins

View full text Add to dashboard Cite

This article reports and analyzes the results of protein complex model accuracy estimation by our methods (DeepUMQA3 and GraphGPSM) in the 15th Critical Assessment of techniques for protein Structure Prediction (CASP15). The new deep learning‐based multimeric complex model accuracy estimation methods are proposed based on the ensemble of three‐level features coupling with deep residual/graph neural networks. For the input multimeric complex model, we describe it from three levels: overall complex features, intra‐monomer features, and inter‐monomer features. We designed an overall ultrafast shape recognition (USR) to characterize the relationship between local residues and the overall complex topology, and an inter‐monomer USR to characterize the relationship between the residues of one monomer and the topology of other monomers. DeepUMQA3 (Group name: GuijunLab‐RocketX) ranked first in the interface residue accuracy estimation of CASP15. The Pearson correlation between the interface residue Local Distance Difference Test (lDDT) predicted by DeepUMQA3 and the real lDDT is 0.570, the only method that exceeds 0.5. Among the top 5 methods, DeepUMQA3 achieved the highest Pearson correlation of lDDT on 25 out of 39 targets. GraphGPSM (Group name: GuijunLab‐PAthreader) has TM‐score Pearson correlations greater than 0.9 on 14 targets, showing a good ability to estimate the overall fold accuracy. The DeepUMQA3 server is available at http://zhanglab-bioinf.com/DeepUMQA/ and the GraphGPSM server is available at http://zhanglab-bioinf.com/GraphGPSM/.

show abstract

Section: Resultsmentioning

confidence: 99%

Estimating protein complex model accuracy based on ultrafast shape recognition and deep learning in CASP15

Liu

et al. 2023

Proteins

View full text Add to dashboard Cite

show abstract

“… a The methods of ModFOLDdock series refer to Edmunds et al (2023) , the method of VoroIF refer to Olechnovič and Venclovas (2023) , the method of MULTICOM_deep refer to Roy et al (2023) , and the other methods refer to the abstract of CASP15 ( https://predictioncenter.org/casp15/doc/CASP15_Abstracts.pdf ). b Pearson and Spearman indicated the correlation between the accuracy of predicted interface residues and the real lDDT/CAD of interface residues.…”

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

DeepUMQA3: a web server for accurate assessment of interface residue accuracy in protein complexes

Liu,

Zhang

2023

Bioinformatics

View full text Add to dashboard Cite

Motivation Model quality assessment is a crucial part of protein structure prediction and a gateway to proper usage of models in biomedical applications. Many methods have been proposed for assessing the quality of structural models of protein monomers, but few methods for evaluating protein complex models. As protein complex structure prediction becomes a new challenge, there is an urgent need for model quality assessment methods that can accurately assess the accuracy of interface residues of complex structures. Results Here, we present DeepUMQA3, a web server for evaluating the accuracy of interface residues of protein complex structures using deep neural networks. For an input complex structure, features are extracted from three levels of overall complex, intra-monomer, and inter-monomer, and a improved deep residual neural network is used to predict per-residue lDDT and interface residue accuracy. DeepUMQA3 ranks first in the blind test of interface residue accuracy estimation in CASP15, with Pearson, Spearman and AUC of 0.564, 0.535 and 0.755 under the lDDT measurement, which are 17.6%, 23.6% and 10.9% higher than the second best method, respectively. DeepUMQA3 can also assess the accuracy of all residues in the entire complex and distinguish high- and low-precision residues. Availability and implementation The web sever of DeepUMQA3 are freely available at http://zhanglab-bioinf.com/DeepUMQA_server/. Supplementary information Supplementary data are available at Bioinformatics online.

show abstract

“…Hence, during the model selection, we relied mostly on the consistency of models with experimental data rather than on predicted ipTM scores. This strategy was based on the previous observations that the predicted contact scores often fail to identify the true models, 99 and experimental verification is usually required to validate and justify the physiological relevance of AF2-predicted conformations of membrane proteins. 100 We found that the V3 version produced better models for TPOR complexes (Figure S13A), but not for other receptors where the results with V2 and V3 versions were rather similar.…”

Section: Modeling Of Signaling Complexes With Alpha-fold-multimer (Afm)mentioning

confidence: 99%

Structural Modeling of Cytokine-Receptor-JAK2 Signaling Complexes Using AlphaFold Multimer

Pogozheva,

Cherepanov,

Park

et al. 2023

J. Chem. Inf. Model.

View full text Add to dashboard Cite

Homodimeric class 1 cytokine receptors include the erythropoietin (EPOR), thrombopoietin (TPOR), granulocyte colony-stimulating factor 3 (CSF3R), growth hormone (GHR), and prolactin receptors (PRLR). These cell-surface single-pass transmembrane (TM) glycoproteins regulate cell growth, proliferation, and differentiation and induce oncogenesis. An active TM signaling complex consists of a receptor homodimer, one or two ligands bound to the receptor extracellular domains, and two molecules of Janus Kinase 2 (JAK2) constitutively associated with the receptor intracellular domains. Although crystal structures of soluble extracellular domains with ligands have been obtained for all of the receptors except TPOR, little is known about the structure and dynamics of the complete TM complexes that activate the downstream JAK-STAT signaling pathway. Three-dimensional models of five human receptor complexes with cytokines and JAK2 were generated here by using AlphaFold Multimer. Given the large size of the complexes (from 3220 to 4074 residues), the modeling required a stepwise assembly from smaller parts, with selection and validation of the models through comparisons with published experimental data. The modeling of active and inactive complexes supports a general activation mechanism that involves ligand binding to a monomeric receptor followed by receptor dimerization and rotational movement of the receptor TM α-helices, causing proximity, dimerization, and activation of associated JAK2 subunits. The binding mode of two eltrombopag molecules to the TM α-helices of the active TPOR dimer was proposed. The models also help elucidate the molecular basis of oncogenic mutations that may involve a noncanonical activation route. Models equilibrated in explicit lipids of the plasma membrane are publicly available.

show abstract

Combining pairwise structural similarity and deep learning interface contact prediction to estimate protein complex model accuracy in CASP15

Cited by 9 publications

References 54 publications

Estimating protein complex model accuracy based on ultrafast shape recognition and deep learning in CASP15

Estimating protein complex model accuracy based on ultrafast shape recognition and deep learning in CASP15

DeepUMQA3: a web server for accurate assessment of interface residue accuracy in protein complexes

Structural Modeling of Cytokine-Receptor-JAK2 Signaling Complexes Using AlphaFold Multimer

Contact Info

Product

Resources

About