MetaPSICOV: combining coevolution methods for accurate prediction of contacts and long range hydrogen bonding in proteins

Abstract

Motivation: Recent developments of statistical techniques to infer direct evolutionary couplings between residue pairs have rendered covariation-based contact prediction a viable means for accurate 3D modelling of proteins, with no information other than the sequence required. To extend the usefulness of contact prediction, we have designed a new meta-predictor (MetaPSICOV) which combines three distinct approaches for inferring covariation signals from multiple sequence alignments, considers a broad range of other sequence-derived features and, uniquely, a range of metrics which describe both the local and global quality of the input multiple sequence alignment. Finally, we use a two-stage predictor, where the second stage filters the output of the first stage. This two-stage predictor is additionally evaluated on its ability to accurately predict the long range network of hydrogen bonds, including correctly assigning the donor and acceptor residues.Results: Using the original PSICOV benchmark set of 150 protein families, MetaPSICOV achieves a mean precision of 0.54 for top-L predicted long range contacts—around 60% higher than PSICOV, and around 40% better than CCMpred. In de novo protein structure prediction using FRAGFOLD, MetaPSICOV is able to improve the TM-scores of models by a median of 0.05 compared with PSICOV. Lastly, for predicting long range hydrogen bonding, MetaPSICOV-HB achieves a precision of 0.69 for the top-L/10 hydrogen bonds compared with just 0.26 for the baseline MetaPSICOV.Availability and implementation: MetaPSICOV is available as a freely available web server at http://bioinf.cs.ucl.ac.uk/MetaPSICOV. Raw data (predicted contact lists and 3D models) and source code can be downloaded from http://bioinf.cs.ucl.ac.uk/downloads/MetaPSICOV.Contact: d.t.jones@ucl.ac.ukSupplementary information: Supplementary data are available at Bioinformatics online.

ConEVA: a toolbox for comprehensive assessment of protein contacts

“…In addition, we also installed a local copy of the tools coevolution based tool CCMpred [2], pure machine-learning based method DNcon [3], and a hybrid method MetaPSICOV [1] to make contact predictions for various data sets including the PSICOV data set of 150 proteins. These contacts along with secondary structures predicted using PSIPRED [33] were used for building models using CONFOLD [8], a fragment-free ab initio method that we recently developed to build 3D models from scratch.…”

“…The success of many protein residue contact prediction methods, in the recent years, has kindled a new hope to solve the long standing problem of ab initio protein structure prediction [1–6]. Consequently, contact-guided ab initio structure prediction has emerged as an important field.…”

“…Consequently, contact-guided ab initio structure prediction has emerged as an important field. When accurately predicted contacts are supplied as input to structure prediction or reconstruction methods, accurate folds can be predicted consistently [1, 7–9]. In general, accurate contacts lead to accurate structural models.…”

“…However, for predicting folds of sequences which do not have homologous templates (hard sequences), the optimal way of utilizing predicted contacts is still an ongoing research. For instance, experiments on true contact reconstruction have suggested that 9 Å or more distance threshold delivers best reconstruction with Cβ atom [10, 11], but the Critical Assessment of Protein Structure Prediction (CASP)’s definition of 8 Å threshold is still widely used to predict contacts [1–3, 6, 12]. Marks et al have even demonstrated successful structure predictions using Cα atoms and 7 Å threshold for defining contacts [13].…”

Identification of residue pairing in interacting β-strands from a predicted residue contact map

“…These methods effectively reduce false positive predictions by globally considering all inter-residue correlations. More recently, methods like MetaPSICOV [19], SAE-DNN [20], DeepConPred [21], NeBcon [22] and RaptorX-Contact [23] integrated sophisticated machine-learning techniques to further enhance the prediction accuracy. In the latest CASP12 competition, RaptorX-Contact achieved the best performance in the category of template-free modeling targets.…”