Predicting protein contact map using evolutionary and physical constraints by integer programming

By Zhiyong Wang, Jinbo Xu. 2013

145 citations recorded by the source corpus.

Motivation: Protein contact map describes the pairwise spatial and functional relationship of residues in a protein and contains key information for protein 3D structure prediction. Although studied extensively, it remains challenging to predict contact map using only sequence information. Most existing methods predict the contact map matrix element-by-element, ignoring correlation among contacts and physical feasibility of the whole-contact map. A couple of recent methods predict contact map by using mutual information, taking into consideration contact correlation and enforcing a sparsity restraint, but these methods demand for a very large number of sequence homologs for the protein under consideration and the resultant contact map may be still physically infeasible. Results: This article presents a novel method PhyCMAP for contact map prediction, integrating both evolutionary and physical restraints by machine learning and integer linear programming. The evolutionary restraints are much more informative than mutual information, and the physical restraints specify more concrete relationship among contacts than the sparsity restraint. As such, our method greatly reduces the solution space of the contact map matrix and, thus, significantly improves prediction accuracy. Experimental results confirm that PhyCMAP outperforms currently popular methods no matter how many sequence homologs are available for the protein under consideration. Availability: http://raptorx.uchicago.edu. Contact: jinboxu@gmail.com

Subfields: Materials Chemistry, Molecular Biology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Materials Chemistry

In Molecular Biology

Citations in the corpus

MRFalign: Protein Homology Detection through Alignment of Markov Random Fields (2014, methodology): “PhyCMAP takes sequence information (including mutual information) as input and predicts the probability of two residues forming a contact, which can be used to indicate the interaction strength of two columns.”
Protein threading using residue co-variation and deep learning (2018, methodology): “As shown in Figure 8, our method can improve alignment accuracy (measured by TM-score) across all Meff values, but the improvement is small when lnðMeff Þ 2, i.e. Meff 7.39.”
Folding membrane proteins by deep transfer learning (2017, background): “Supervised learning predicts contacts from a variety of information, e.g., SVMSEQ(Wu and Zhang, 2008), CMAPpro (Di Lena et al., 2012), PconsC2 (Skwark et al., 2014) and PconsC3 (Skwark et al., 2016), MetaPSICOV (Jones et al., 2015), PhyCMAP (Wang and Xu, 2013) and CoinDCA-NN (Ma et al., 2015).”
Improved Contact Predictions Using the Recognition of Protein Like Contact Patterns (2014, methodology): “PconsC2 performs better than earlier predictors In addition we compare PconsC2 to a set of alternative contact prediction methods, including mutual information (MI) and MI with phylogenetic correction (CMIr), and two recent machine learning based methods, PhyCMAP and CMAPpro.”
Protein contact prediction by integrating joint evolutionary coupling analysis and supervised learning (2013, methodology): “Similar to (Marks et al., 2011; Wang and Xu, 2013), we calculate the number of non-redundant sequence homologs in a family (or MSA) by Meff 1⁄4 P i1= P jsi;j where i and j are sequence indexes and si,j is a binary variable indicating if two sequences are similar or not.”
Accurate prediction of protein contact maps by coupling residual two-dimensional bidirectional long short-term memory with convolutional neural networks (2018, methodology): “While these methods are useful for predicting long-range contacts in proteins with a high number of sequence homologues, their accuracy is poor if the number of homologues is low (Wang and Xu, 2013).”
A new ensemble coevolution system for detecting HIV-1 protein coevolution (2015, methodology): “We used the contact map for our evaluation purpose because it is still one of the most widely-used standard measurements in the coevolution field, to compare different sequence-based methods [25,48,50] .”
Deducing high-accuracy protein contact-maps from a triplet of coevolutionary matrices through deep residual convolutional networks (2020, background): “Early attempts utilized support vector machines (SVMs) [17,18], random forests (RFs) [13,19], artificial neural networks (NNs) [20–23] etc.”
Protein Homology Detection Through Alignment of Markov Random Fields (2015, methodology): “We can use other contact prediction programs such as PhyCMAP [4] which integrates both residue col-evolution information, PSI-BLAST sequence profile and others to predict the probability of two residues in contact.”
Knowledge-based machine learning methods for macromolecular 3D structure prediction (2016, background): “This PSSM feature is proved very helpful in many sequence related research (Jones, 1999; Källberg, et al., 2012; Ma, et al., 2014; Peng and Xu, 2010; Peng and Xu, 2011; Söding, 2005; Wang and Xu, 2013; Wang, et al., 2011).”
Analysis of deep learning methods for blind protein contact prediction in CASP12 (2017, background): “Supervised learning such as PhyCMAP[9], DNCON[10] and SVMSEQ[11] predicts contacts using a variety of protein features, on average outperforming pure co-evolution methods on proteins with few sequence homologs.”
UniCon3D: de novo protein structure prediction using united-residue conformational search via stepwise, probabilistic sampling (2016, background): “…et al., 2011; Seemayer et al., 2014; Skwark et al., 2013) or machine learning based methods (Cheng and Baldi, 2007; Eickholt and Cheng, 2012; Tegge et al., 2009; Wang and Xu, 2013) or a combination of both covariation techniques and machine learning (Jones et al., 2015; Kosciolek and Jones, 2015).”
FreeContact: fast and free software for protein contact prediction from residue co-evolution (2014, background): “PSICOV [11], plmDCA [12], PconsC [13] and PhyCMAP [14-16], and EVfold_membrane [7]).”
Predicting membrane protein contacts from non-membrane proteins by deep transfer learning (2017, unknown): “Supervised learning predicts contacts from a variety of information, e.g., SVMSEQ[13], CMAPpro [14], PconsC2 [11] and PconsC3 [15], MetaPSICOV [16], PhyCMAP [17] and CoinDCA-NN [12].”
Predicting accurate contacts in thousands of Pfam domain families using PconsC3 (2017, methodology): “The DCA methods were directly run on the alignments described above, whereas PhyCMAP runs its own workflow and thus uses its own alignment (Wang and Xu, 2013).”
Fast pseudolikelihood maximization for direct-coupling analysis of protein structure from many homologous amino-acid sequences (2014, methodology): “Numerous freshly conceived methods expand the concepts and applicability of DCA in various directions [30, 31, 32, 33, 34, 35, 36, 37, 38].”