Protein Contact Prediction by Integrating Joint Evolutionary Coupling Analysis and Supervised Learning

Ma, Jianzhu; Wang, Sheng; Wang, Zhiyong; Xu, Jinbo

doi:10.1007/978-3-319-16706-0_21

Cited by 54 publications

(84 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We evaluate the accuracy of the top L/k ( k = 10, 5, 2, 1) predicted contacts where L is protein sequence length [10]. We define that a contact is short-, medium- and long-range when the sequence distance of the two residues in a contact falls into [6, 11], [12, 23], and ≥24, respectively.…”

Section: Resultsmentioning

confidence: 99%

“…Therefore, contact prediction and contact-assisted protein folding has recently gained much attention in the community. However, for many proteins especially those without many sequence homologs, the predicted contacts by the state-of-the-art predictors such as CCMpred [4], PSICOV [5], Evfold [6], plmDCA[7], Gremlin[8], MetaPSICOV [9] and CoinDCA [10] are still of low quality and insufficient for accurate contact-assisted protein folding [11, 12]. This motivates us to develop a better contact prediction method, especially for proteins without a large number of sequence homologs.…”

Section: Introductionmentioning

confidence: 99%

“…ECA predicts contacts by identifying co-evolved residues in a protein, such as EVfold [6], PSICOV [5], CCMpred [4], Gremlin [8], plmDCA and others [14–16]. However, DCA usually needs a large number of sequence homologs to be effective [10, 17]. Supervised machine learning predicts contacts from a variety of information, e.g., SVMSEQ [18], CMAPpro [13], PconsC2 [17], MetaPSICOV [9], PhyCMAP [19] and CoinDCA-NN [10].…”

Section: Introductionmentioning

confidence: 99%

“…However, DCA usually needs a large number of sequence homologs to be effective [10, 17]. Supervised machine learning predicts contacts from a variety of information, e.g., SVMSEQ [18], CMAPpro [13], PconsC2 [17], MetaPSICOV [9], PhyCMAP [19] and CoinDCA-NN [10]. Meanwhile, PconsC2 uses a 5-layer supervised learning architecture [17]; CoinDCA-NN and MetaPSICOV employ a 2-layer neural network [9].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Accurate De Novo Prediction of Protein Contact Map by Ultra-Deep Learning Model

et al. 2017

Self Cite

View full text Add to dashboard Cite

MotivationProtein contacts contain key information for the understanding of protein structure and function and thus, contact prediction from sequence is an important problem. Recently exciting progress has been made on this problem, but the predicted contacts for proteins without many sequence homologs is still of low quality and not very useful for de novo structure prediction.MethodThis paper presents a new deep learning method that predicts contacts by integrating both evolutionary coupling (EC) and sequence conservation information through an ultra-deep neural network formed by two deep residual neural networks. The first residual network conducts a series of 1-dimensional convolutional transformation of sequential features; the second residual network conducts a series of 2-dimensional convolutional transformation of pairwise information including output of the first residual network, EC information and pairwise potential. By using very deep residual networks, we can accurately model contact occurrence patterns and complex sequence-structure relationship and thus, obtain higher-quality contact prediction regardless of how many sequence homologs are available for proteins in question.ResultsOur method greatly outperforms existing methods and leads to much more accurate contact-assisted folding. Tested on 105 CASP11 targets, 76 past CAMEO hard targets, and 398 membrane proteins, the average top L long-range prediction accuracy obtained by our method, one representative EC method CCMpred and the CASP11 winner MetaPSICOV is 0.47, 0.21 and 0.30, respectively; the average top L/10 long-range accuracy of our method, CCMpred and MetaPSICOV is 0.77, 0.47 and 0.59, respectively. Ab initio folding using our predicted contacts as restraints but without any force fields can yield correct folds (i.e., TMscore>0.6) for 203 of the 579 test proteins, while that using MetaPSICOV- and CCMpred-predicted contacts can do so for only 79 and 62 of them, respectively. Our contact-assisted models also have much better quality than template-based models especially for membrane proteins. The 3D models built from our contact prediction have TMscore>0.5 for 208 of the 398 membrane proteins, while those from homology modeling have TMscore>0.5 for only 10 of them. Further, even if trained mostly by soluble proteins, our deep learning method works very well on membrane proteins. In the recent blind CAMEO benchmark, our fully-automated web server implementing this method successfully folded 6 targets with a new fold and only 0.3L-2.3L effective sequence homologs, including one β protein of 182 residues, one α+β protein of 125 residues, one α protein of 140 residues, one α protein of 217 residues, one α/β of 260 residues and one α protein of 462 residues. Our method also achieved the highest F1 score on free-modeling targets in the latest CASP (Critical Assessment of Structure Prediction), although it was not fully implemented back then.Availabilityhttp://raptorx.uchicago.edu/ContactMap/

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Accurate De Novo Prediction of Protein Contact Map by Ultra-Deep Learning Model

et al. 2017

Self Cite

View full text Add to dashboard Cite

show abstract

“…However, soluble proteins with that many sequence homologs likely have similar solved structures in protein database (PDB) and thus, may be modeled by template‐based methods. For proteins with few sequence homologs, pure coevolution methods such as CCMpred, PSICOV, Evfold, Gremlin, and CoinDCA do not fare well and their predictions are not very helpful to ab initio folding. Supervised learning such as PhyCMAP, DNCON, and SVMSEQ predicts contacts using a variety of protein features, on average outperforming pure coevolution methods on proteins with few sequence homologs.…”

Section: Introductionmentioning

confidence: 99%

Analysis of deep learning methods for blind protein contact prediction in CASP12

2017

Self Cite

View full text Add to dashboard Cite

Here we present the results of protein contact prediction achieved in CASP12 by our RaptorX-Contact server, which is an early implementation of our deep learning method for contact prediction. On a set of 38 free-modeling target domains with a median family size of around 58 effective sequences, our server obtained an average top L/5 long- and medium-range contact accuracy of 47% and 44%, respectively (L=length). A more advanced implementation has an average accuracy of 59% and 57%, respectively. Our deep learning method formulates contact prediction as a pixel-level labeling problem and simultaneously predicts all residue pairs of a protein using a combination of two deep residual neural networks, taking as input the residue conservation information, predicted secondary structure and solvent accessibility, contact potential, and co-evolution information. Our approach differs from existing methods mainly in (1) formulating contact prediction as a pixel-level image labeling problem instead of an image-level classification problem; (2) simultaneously predicting all contacts of an individual protein to make effective use of contact occurrence patterns; and (3) integrating both 1D and 2D deep convolutional neural networks to effectively learn complex sequence-structure relationship including high-order residue correlation. This paper discusses the RaptorX-Contact pipeline, both contact prediction and contact-based folding results, and finally the strength and weakness of our method.

show abstract

Blind testing of cross‐linking/mass spectrometry hybrid methods in CASP11

et al. 2016

View full text Add to dashboard Cite

Hybrid approaches combine computational methods with experimental data. The information contained in the experimental data can be leveraged to probe the structure of proteins otherwise elusive to computational methods. Compared with computational methods, the structures produced by hybrid methods exhibit some degree of experimental validation. In spite of these advantages, most hybrid methods have not yet been validated in blind tests, hampering their development. Here, we describe the first blind test of a specific cross‐link based hybrid method in CASP. This blind test was coordinated by the CASP organizers and utilized a novel, high‐density cross‐linking/mass‐spectrometry (CLMS) approach that is able to collect high‐density CLMS data in a matter of days. This experimental protocol was developed in the Rappsilber laboratory. This approach exploits the chemistry of a highly reactive, photoactivatable cross‐linker to produce an order of magnitude more cross‐links than homobifunctional cross‐linkers. The Rappsilber laboratory generated experimental CLMS data based on this protocol, submitted the data to the CASP organizers which then released this data to the CASP11 prediction groups in a separate, CLMS assisted modeling experiment. We did not observe a clear improvement of assisted models, presumably because the properties of the CLMS data—uncertainty in cross‐link identification and residue‐residue assignment, and uneven distribution over the protein—were largely unknown to the prediction groups and their approaches were not yet tailored to this kind of data. We also suggest modifications to the CLMS‐CASP experiment and discuss the importance of rigorous blind testing in the development of hybrid methods. Proteins 2016; 84(Suppl 1):152–163. © 2016 The Authors Proteins: Structure, Function, and Bioinformatics Published by Wiley Periodicals, Inc.

show abstract

Protein Contact Prediction by Integrating Joint Evolutionary Coupling Analysis and Supervised Learning

Cited by 54 publications

References 45 publications

Accurate De Novo Prediction of Protein Contact Map by Ultra-Deep Learning Model

Accurate De Novo Prediction of Protein Contact Map by Ultra-Deep Learning Model

Analysis of deep learning methods for blind protein contact prediction in CASP12

Blind testing of cross‐linking/mass spectrometry hybrid methods in CASP11

Contact Info

Product

Resources

About