COMSAT: Residue contact prediction of transmembrane proteins based on support vector machines and mixed integer linear programming

Abstract

In this article, we present COMSAT, a hybrid framework for residue contact prediction of transmembrane (TM) proteins, integrating a support vector machine (SVM) method and a mixed integer linear programming (MILP) method. COMSAT consists of two modules: COMSAT_SVM which is trained mainly on position-specific scoring matrix features, and COMSAT_MILP which is an ab initio method based on optimization models. Contacts predicted by the SVM model are ranked by SVM confidence scores, and a threshold is trained to improve the reliability of the predicted contacts. For TM proteins with no contacts above the threshold, COMSAT_MILP is used. The proposed hybrid contact prediction scheme was tested on two independent TM protein sets based on the contact definition of 14 Å between Cα-Cα atoms. First, using a rigorous leave-one-protein-out cross validation on the training set of 90 TM proteins, an accuracy of 66.8%, a coverage of 12.3%, a specificity of 99.3% and a Matthews' correlation coefficient (MCC) of 0.184 were obtained for residue pairs that are at least six amino acids apart. Second, when tested on a test set of 87 TM proteins, the proposed method showed a prediction accuracy of 64.5%, a coverage of 5.3%, a specificity of 99.4% and a MCC of 0.106. COMSAT shows satisfactory results when compared with 12 other state-of-the-art predictors, and is more robust in terms of prediction accuracy as the length and complexity of TM protein increase. COMSAT is freely accessible at http://hpcc.siat.ac.cn/COMSAT/.

ConEVA: a toolbox for comprehensive assessment of protein contacts

“…Using the protein 1JWQ, Vassura et al show how some structures cannot be folded with distance thresholds below 16 Å [10]. Zhang et al report folding 90 transmembrane proteins at 14 Å cut-off [17]. Furthermore, in these works, no common agreement is found on the optimal number of contacts (or a range) needed for accurate reconstruction.…”

Folding Membrane Proteins by Deep Transfer Learning

“…We have talked to Prof. David Jones, who have developed both MP-specific tool MEMPACK and generic contact prediction tools PSICOV and MetaPSICOV and informed us that MEMPACK is not as good as MetaPSICOV. A recent paper (Zhang et al, 2016a) proposed a new MP-specific tool COMSAT, compared 12 MP-specific and MP-independent contact prediction tools and showed that MP-specific tools have no significant advantage over the best MP-independent tools. To further verify this, we have tested our deep learning model (trained by non-MPs only) on the 87 membrane proteins tested in the COMSAT paper.…”

“…There are some contact prediction methods specifically developed for MPs. They employ some MP-specific features and are trained from a limited number of MPs, such as TMHcon (Fuchs et al, 2009), MEMPACK (Nugent and Jones, 2010), TMhit (Lo et al, 2009), TMhhcp (Wang et al, 2011), MemBrain (Yang et al, 2013), COMSAT (Zhang et al, 2016a) and OMPcontact (Zhang et al, 2016b). McAllister and Floudas (McAllister and Floudas, 2008) proposed a mixed integer programming method for MP contact prediction by optimizing an energy function subject to a set of physical constraints.…”

IMPContact: An Interhelical Residue Contact Prediction Method

“…For the purpose of IHRC prediction, various sequenceor structure-based features and machine learning algorithms were utilized. Neural network (NN), support vector machines (SVMs), and random forest (RF) algorithms were, respectively, proven to be effective in predicting IHRCs in TMHcon [13], TMhit [14], MEMPACK [15], TMhhcp [4], COMSAT [16], and MemConP [17]. It is convincing that the coevolution relationship highly improved the prediction accuracy used in TMHcon [13], MEMPACK [15], and MemConP [17].…”