Assessment Of Protein Structure Prediction Research Articles

Assessment of the model refinement category in CASP12.

We here report on the assessment of the model refinement predictions submitted to the 12th Experiment on the Critical Assessment of Protein Structure Prediction (CASP12). This is the fifth refinement experiment since CASP8 (2008) and, as with the previous experiments, the predictors were invited to refine selected server models received in the regular (nonrefinement) stage of the CASP experiment. We assessed the submitted models using a combination of standard CASP measures. The coefficients for the linear combination of Z-scores (the CASP12 score) have been obtained by a machine learning algorithm trained on the results of visual inspection. We identified eight groups that improve both the backbone conformation and the side chain positioning for the majority of targets. Albeit the top methods adopted distinctively different approaches, their overall performance was almost indistinguishable, with each of them excelling in different scores or target subsets. What is more, there were a few novel approaches that, while doing worse than average in most cases, provided the best refinements for a few targets, showing significant latitude for further innovation in the field.

A comprehensive review and comparison of existing computational methods for intrinsically disordered protein and region prediction

Intrinsically disordered proteins and regions are widely distributed in proteins, which are associated with many biological processes and diseases. Accurate prediction of intrinsically disordered proteins and regions is critical for both basic research (such as protein structure and function prediction) and practical applications (such as drug development). During the past decades, many computational approaches have been proposed, which have greatly facilitated the development of this important field. Therefore, a comprehensive and updated review is highly required. In this regard, we give a review on the computational methods for intrinsically disordered protein and region prediction, especially focusing on the recent development in this field. These computational approaches are divided into four categories based on their methodologies, including physicochemical-based method, machine-learning-based method, template-based method and meta method. Furthermore, their advantages and disadvantages are also discussed. The performance of 40 state-of-the-art predictors is directly compared on the target proteins in the task of disordered region prediction in the 10th Critical Assessment of protein Structure Prediction. A more comprehensive performance comparison of 45 different predictors is conducted based on seven widely used benchmark data sets. Finally, some open problems and perspectives are discussed.

Modeling of protein complexes in CAPRI Round 37 using template-based approach combined with model selection.

We participated in Round 37 of the Critical Assessment of PRediction of Interactions (CAPRI), held jointly with the 12th edition of the Critical Assessment of protein Structure Prediction (CASP12), having two major objectives. First, we intended to test the utility of our PPI3D web server in finding and selecting templates for comparative modeling of structures of protein complexes. Our second aim was to evaluate the ability of our model accuracy estimation method VoroMQA to score and rank structural models for protein-protein interactions. Using sequence search in PPI3D and HHpred servers we identified multimeric templates for 7 of 11 CAPRI targets, and models of at least acceptable quality were constructed for 6 of them. The clustering and visual analysis features implemented in the PPI3D software were instrumental in detecting alternative protein-protein interaction interfaces among the identified templates. When a single binding mode was observed for homologous proteins, the structural modeling of the protein complex was fairly straightforward, whereas choosing the correct interaction template from several alternatives turned out to be a difficult task requiring manual intervention. The combination of full structure and interaction interface VoroMQA scores effectively ranked structural models of protein complexes and selected models of better quality from the CAPRI Scoring sets. The overall results show possible uses of PPI3D and VoroMQA in structural modeling of protein-protein interactions and suggest ways for further improvements of both methods.

GASS-WEB: a web server for identifying enzyme active sites based on genetic algorithms.

Enzyme active sites are important and conserved functional regions of proteins whose identification can be an invaluable step toward protein function prediction. Most of the existing methods for this task are based on active site similarity and present limitations including performing only exact matches on template residues, template size restraints, despite not being capable of finding inter-domain active sites. To fill this gap, we proposed GASS-WEB, a user-friendly web server that uses GASS (Genetic Active Site Search), a method based on an evolutionary algorithm to search for similar active sites in proteins. GASS-WEB can be used under two different scenarios: (i) given a protein of interest, to match a set of specific active site templates; or (ii) given an active site template, looking for it in a database of protein structures. The method has shown to be very effective on a range of experiments and was able to correctly identify >90% of the catalogued active sites from the Catalytic Site Atlas. It also managed to achieve a Matthew correlation coefficient of 0.63 using the Critical Assessment of protein Structure Prediction (CASP 10) dataset. In our analysis, GASS was ranking fourth among 18 methods. GASS-WEB is freely available at http://gass.unifei.edu.br/.

Reduced Fragment Diversity for Alpha and Alpha-Beta Protein Structure Prediction using Rosetta.

Protein structure prediction is considered a main challenge in computational biology. The biannual international competition, Critical Assessment of protein Structure Prediction (CASP), has shown in its eleventh experiment that free modelling target predictions are still beyond reliable accuracy, therefore, much effort should be made to improve ab initio methods. Arguably, Rosetta is considered as the most competitive method when it comes to targets with no homologues. Relying on fragments of length 9 and 3 from known structures, Rosetta creates putative structures by assembling candidate fragments. Generally, the structure with the lowest energy score, also known as first model, is chosen to be the "predicted one". A thorough study has been conducted on the role and diversity of 3-mers involved in Rosetta's model "refinement" phase. Usage of the standard number of 3-mers - i.e. 200 - has been shown to degrade alpha and alpha-beta protein conformations initially achieved by assembling 9-mers. Therefore, a new prediction pipeline is proposed for Rosetta where the "refinement" phase is customised according to a target's structural class prediction. Over 8% improvement in terms of first model structure accuracy is reported for alpha and alpha-beta classes when decreasing the number of 3- mers.

Blind testing cross-linking/mass spectrometry under the auspices of the 11th critical assessment of methods of protein structure prediction (CASP11).

Determining the structure of a protein by any method requires various contributions from experimental and computational sides. In a recent study, high-density cross-linking/mass spectrometry (HD-CLMS) data in combination with ab initio structure prediction determined the structure of human serum albumin (HSA) domains, with an RMSD to X-ray structure of up to 2.5 Å, or 3.4 Å in the context of blood serum. This paper reports the blind test on the readiness of this technology through the help of Critical Assessment of protein Structure Prediction (CASP). We identified between 201-381 unique residue pairs at an estimated 5% FDR (at link level albeit with missing site assignment precision evaluation), for four target proteins. HD-CLMS proved reliable once crystal structures were released. However, improvements in structure prediction using cross-link data were slight. We identified two reasons for this. Spread of cross-links along the protein sequence and the tightness of the spatial constraints must be improved. However, for the selected targets even ideal contact data derived from crystal structures did not allow modellers to arrive at the observed structure. Consequently, the progress of HD-CLMS in conjunction with computational modeling methods as a structure determination method, depends on advances on both arms of this hybrid approach.

DeepQA: improving the estimation of single protein model quality with deep belief networks.

BackgroundProtein quality assessment (QA) useful for ranking and selecting protein models has long been viewed as one of the major challenges for protein tertiary structure prediction. Especially, estimating the quality of a single protein model, which is important for selecting a few good models out of a large model pool consisting of mostly low-quality models, is still a largely unsolved problem.ResultsWe introduce a novel single-model quality assessment method DeepQA based on deep belief network that utilizes a number of selected features describing the quality of a model from different perspectives, such as energy, physio-chemical characteristics, and structural information. The deep belief network is trained on several large datasets consisting of models from the Critical Assessment of Protein Structure Prediction (CASP) experiments, several publicly available datasets, and models generated by our in-house ab initio method. Our experiments demonstrate that deep belief network has better performance compared to Support Vector Machines and Neural Networks on the protein model quality assessment problem, and our method DeepQA achieves the state-of-the-art performance on CASP11 dataset. It also outperformed two well-established methods in selecting good outlier models from a large set of models of mostly low quality generated by ab initio modeling methods.ConclusionDeepQA is a useful deep learning tool for protein single model quality assessment and protein structure prediction. The source code, executable, document and training/test datasets of DeepQA for Linux is freely available to non-commercial users at http://cactus.rnet.missouri.edu/DeepQA/.Electronic supplementary materialThe online version of this article (doi:10.1186/s12859-016-1405-y) contains supplementary material, which is available to authorized users.

Assessment of template-based modeling of protein structure in CASP11.

We present the assessment of predictions submitted in the template-based modeling (TBM) category of CASP11 (Critical Assessment of Protein Structure Prediction). Model quality was judged on the basis of global and local measures of accuracy on all atoms including side chains. The top groups on 39 human-server targets based on model 1 predictions were LEER, Zhang, LEE, MULTICOM, and Zhang-Server. The top groups on 81 targets by server groups based on model 1 predictions were Zhang-Server, nns, BAKER-ROSETTASERVER, QUARK, and myprotein-me. In CASP11, the best models for most targets were equal to or better than the best template available in the Protein Data Bank, even for targets with poor templates. The overall performance in CASP11 is similar to the performance of predictors in CASP10 with slightly better performance on the hardest targets. For most targets, assessment measures exhibited bimodal probability density distributions. Multi-dimensional scaling of an RMSD matrix for each target typically revealed a single cluster with models similar to the target structure, with a mode in the GDT-TS density between 40 and 90, and a wide distribution of models highly divergent from each other and from the experimental structure, with density mode at a GDT-TS value of ∼20. The models in this peak in the density were either compact models with entirely the wrong fold, or highly non-compact models. The results argue for a density-driven approach in future CASP TBM assessments that accounts for the bimodal nature of these distributions instead of Z scores, which assume a unimodal, Gaussian distribution. Proteins 2016; 84(Suppl 1):200-220. © 2016 Wiley Periodicals, Inc.

Assessment of refinement of template-based models in CASP11.

CASP11 (the 11th Meeting on the Critical Assessment of Protein Structure Prediction) ran a blind experiment in the refinement of protein structure predictions, the fourth such experiment since CASP8. As with the previous experiments, the predictors were provided with one starting structure from the server models of each of a selected set of template-based modeling targets and asked to refine the coordinates of the starting structure toward native. We assessed the refined structures with the Z-scores of the standard CASP measures, which compare the model-target similarities of the models from all the predictors. Furthermore, we assessed the refined structures with "relative measures," which compare the improvement in accuracy of each model with respect to the starting structure. The latter provides an assessment of the extent to which each predictor group is able to improve the starting structures toward native. We utilized heat maps to display improvements in the Calpha-Calpha distance matrix for each model. The heat maps labeled with each element of secondary structure helped us to identify regions of refinement toward native in each model. Most positively scoring models show modest improvements in multiple regions of the structure, while in some models we were able to identify significant repositioning of N/C-terminal segments and internal elements of secondary structure. The best groups were able to improve more than 70% of the targets from the starting models, and by an average of 3-5% in the standard CASP measures. Proteins 2016; 84(Suppl 1):260-281. © 2016 Wiley Periodicals, Inc.

UniCon3D: de novo protein structure prediction using united-residue conformational search via stepwise, probabilistic sampling.

Motivation: Recent experimental studies have suggested that proteins fold via stepwise assembly of structural units named ‘foldons’ through the process of sequential stabilization. Alongside, latest developments on computational side based on probabilistic modeling have shown promising direction to perform de novo protein conformational sampling from continuous space. However, existing computational approaches for de novo protein structure prediction often randomly sample protein conformational space as opposed to experimentally suggested stepwise sampling.Results: Here, we develop a novel generative, probabilistic model that simultaneously captures local structural preferences of backbone and side chain conformational space of polypeptide chains in a united-residue representation and performs experimentally motivated conditional conformational sampling via stepwise synthesis and assembly of foldon units that minimizes a composite physics and knowledge-based energy function for de novo protein structure prediction. The proposed method, UniCon3D, has been found to (i) sample lower energy conformations with higher accuracy than traditional random sampling in a small benchmark of 6 proteins; (ii) perform comparably with the top five automated methods on 30 difficult target domains from the 11th Critical Assessment of Protein Structure Prediction (CASP) experiment and on 15 difficult target domains from the 10th CASP experiment; and (iii) outperform two state-of-the-art approaches and a baseline counterpart of UniCon3D that performs traditional random sampling for protein modeling aided by predicted residue-residue contacts on 45 targets from the 10th edition of CASP.Availability and Implementation: Source code, executable versions, manuals and example data of UniCon3D for Linux and OSX are freely available to non-commercial users at http://sysbio.rnet.missouri.edu/UniCon3D/.Contact: chengji@missouri.eduSupplementary information: Supplementary data are available at Bioinformatics online.

The Maximum Entropy Fallacy Redux?

Protein structure can be predicted in silico given sufficiently good templates, as demonstrated in successive installments of the biannual Critical Assessment of protein Structure Prediction (CASP) competition [1]. Since the number of known protein sequences currently increases much faster than the number of known protein structures, and is likely to continue to do so in the foreseeable future, reliable ab initio protein structure prediction, without recourse to templates, would be highly desirable. It is currently not possible to achieve sufficient sampling in unrestrained folding to achieve predictions close to the native structure. Recently, very important progress has been made on the restricted problem of predicting spatial amino acid contacts in proteins from many homologous sequences [2]. While it is not yet clear if these techniques, collectively known as direct coupling analysis (DCA), can be leveraged to systematically predict full protein structures, preliminary results indicate that this may be the case [3,4]. DCA has also been used with success for several related problems, such as predicting structures of protein complexes [4] or alternative protein conformations [5]. The central ingredient in DCA is to learn generative probabilistic models from a set of homologous protein sequences. These models are chosen from an exponential family with linear and quadratic interactions, commonly referred to as Potts models (see Eq 1) [6]. In the literature, this procedure has been motivated by maximum entropy arguments [7–10]. In this Perspective, I will point out that these arguments are mistaken and that the successes of DCA can have nothing to do with maximum entropy. To the contrary, maximum entropy hides the real nature and questions raised by DCA and is thus an obstacle to progress. In addition, maximum entropy has a long and contested history in statistical physics, the field in which it was first introduced [11,12]. Definite and precise results derived in the last decade and a half have here conclusively falsified maximum entropy. Appeals to maximum entropy are therefore prejudicial to a more general acceptance and adoption of DCA.

CASP11--An Evaluation of a Modular BCL::Fold-Based Protein Structure Prediction Pipeline.

In silico prediction of a protein’s tertiary structure remains an unsolved problem. The community-wide Critical Assessment of Protein Structure Prediction (CASP) experiment provides a double-blind study to evaluate improvements in protein structure prediction algorithms. We developed a protein structure prediction pipeline employing a three-stage approach, consisting of low-resolution topology search, high-resolution refinement, and molecular dynamics simulation to predict the tertiary structure of proteins from the primary structure alone or including distance restraints either from predicted residue-residue contacts, nuclear magnetic resonance (NMR) nuclear overhauser effect (NOE) experiments, or mass spectroscopy (MS) cross-linking (XL) data. The protein structure prediction pipeline was evaluated in the CASP11 experiment on twenty regular protein targets as well as thirty-three ‘assisted’ protein targets, which also had distance restraints available. Although the low-resolution topology search module was able to sample models with a global distance test total score (GDT_TS) value greater than 30% for twelve out of twenty proteins, frequently it was not possible to select the most accurate models for refinement, resulting in a general decay of model quality over the course of the prediction pipeline. In this study, we provide a detailed overall analysis, study one target protein in more detail as it travels through the protein structure prediction pipeline, and evaluate the impact of limited experimental data.

Assessment of CASP11 contact-assisted predictions.

We present an overview of contact-assisted predictions in the eleventh round of critical assessment of protein structure prediction (CASP11), which included four categories: predicted contacts (Tp), correct contacts (Tc), simulated sparse NMR contacts (Ts), and cross-linking contacts (Tx). Comparison of assisted to unassisted model quality highlighted a relatively poor overall performance in CASP11 using predicted Tp and crosslinked Tx contact information. However, average model quality significantly improved in the correct Tc and simulated NMR Ts categories for most targets, where maximum improvement of unassisted models reached an impressive 70 GDT_TS. Comparison of the performance in the correct Tc category to CASP10 suggested the improvement in CASP11 model quality originated from an increased number of provided contacts per target. Group rankings based on a combination of scores used in the CASP11 free modeling (FM) assessment for each category highlight four top-performing groups, with three from the Lee lab and one from the Baker lab. We used the overall performance of these groups in each category to develop hypotheses for their relative outperformance in the correct Tc and simulated NMR Ts categories, which stemmed from the fraction of correct contacts provided (correct Tc category) and a reduced fraction of correct contacts offset by an increased coverage of the correct contacts (simulated NMR Ts category). Proteins 2016; 84(Suppl 1):164-180. © 2016 Wiley Periodicals, Inc.

Protein structure prediction

Predicting 3D structure of proteins from the amino acid sequences is one of the most important unsolved problems in computational biology and biophysics. This review article attempts to introduce the most recent effort and progress on this problem. After a brief introduction of the background and basic concepts involved in protein structure prediction, we went through the specific steps that have been taken by most typical structural modeling approaches, including fold recognition, model initialization, conformational search, model selection, and atomic-level structure refinement. Several representative structure prediction methods were introduced in detail, including those from both template-based modeling and ab initio folding approaches. Finally, we overview the results shown in the community-wide Critical Assessment of protein Structure Prediction (CASP) experiments that have been developed for benchmarking the state of the art of the field.

Some of the most interesting CASP11 targets through the eyes of their authors.

ABSTRACTThe Critical Assessment of protein Structure Prediction (CASP) experiment would not have been possible without the prediction targets provided by the experimental structural biology community. In this article, selected crystallographers providing targets for the CASP11 experiment discuss the functional and biological significance of the target proteins, highlight their most interesting structural features, and assess whether these features were correctly reproduced in the predictions submitted to CASP11. Proteins 2016; 84(Suppl 1):34–50. © 2015 The Authors. Proteins: Structure, Function, and Bioinformatics Published by Wiley Periodicals, Inc.

FALCON@home: a high-throughput protein structure prediction server based on remote homologue recognition.

The protein structure prediction approaches can be categorized into template-based modeling (including homology modeling and threading) and free modeling. However, the existing threading tools perform poorly on remote homologous proteins. Thus, improving fold recognition for remote homologous proteins remains a challenge. Besides, the proteome-wide structure prediction poses another challenge of increasing prediction throughput. In this study, we presented FALCON@home as a protein structure prediction server focusing on remote homologue identification. The design of FALCON@home is based on the observation that a structural template, especially for remote homologous proteins, consists of conserved regions interweaved with highly variable regions. The highly variable regions lead to vague alignments in threading approaches. Thus, FALCON@home first extracts conserved regions from each template and then aligns a query protein with conserved regions only rather than the full-length template directly. This helps avoid the vague alignments rooted in highly variable regions, improving remote homologue identification. We implemented FALCON@home using the Berkeley Open Infrastructure of Network Computing (BOINC) volunteer computing protocol. With computation power donated from over 20,000 volunteer CPUs, FALCON@home shows a throughput as high as processing of over 1000 proteins per day. In the Critical Assessment of protein Structure Prediction (CASP11), the FALCON@home-based prediction was ranked the 12th in the template-based modeling category. As an application, the structures of 880 mouse mitochondria proteins were predicted, which revealed the significant correlation between protein half-lives and protein structural factors. FALCON@home is freely available at http://protein.ict.ac.cn/FALCON/. shuaicli@cityu.edu.hk, dbu@ict.ac.cn Supplementary data are available at Bioinformatics online.

General overview on structure prediction of twilight-zone proteins.

Protein structure prediction from amino acid sequence has been one of the most challenging aspects in computational structural biology despite significant progress in recent years showed by critical assessment of protein structure prediction (CASP) experiments. When experimentally determined structures are unavailable, the predictive structures may serve as starting points to study a protein. If the target protein consists of homologous region, high-resolution (typically <1.5 Å) model can be built via comparative modelling. However, when confronted with low sequence similarity of the target protein (also known as twilight-zone protein, sequence identity with available templates is less than 30 %), the protein structure prediction has to be initiated from scratch. Traditionally, twilight-zone proteins can be predicted via threading or ab initio method. Based on the current trend, combination of different methods brings an improved success in the prediction of twilight-zone proteins. In this mini review, the methods, progresses and challenges for the prediction of twilight-zone proteins were discussed.

Assessment of the utility of contact-based restraints in accelerating the prediction of protein structure using molecular dynamics simulations.

Molecular dynamics (MD) simulation is a well-established tool for the computational study of protein structure and dynamics, but its application to the important problem of protein structure prediction remains challenging, in part because extremely long timescales can be required to reach the native structure. Here, we examine the extent to which the use of low-resolution information in the form of residue-residue contacts, which can often be inferred from bioinformatics or experimental studies, can accelerate the determination of protein structure in simulation. We incorporated sets of 62, 31, or 15 contact-based restraints in MD simulations of ubiquitin, a benchmark system known to fold to the native state on the millisecond timescale in unrestrained simulations. One-third of the restrained simulations folded to the native state within a few tens of microseconds-a speedup of over an order of magnitude compared with unrestrained simulations and a demonstration of the potential for limited amounts of structural information to accelerate structure determination. Almost all of the remaining ubiquitin simulations reached near-native conformations within a few tens of microseconds, but remained trapped there, apparently due to the restraints. We discuss potential methodological improvements that would facilitate escape from these near-native traps and allow more simulations to quickly reach the native state. Finally, using a target from the Critical Assessment of protein Structure Prediction (CASP) experiment, we show that distance restraints can improve simulation accuracy: In our simulations, restraints stabilized the native state of the protein, enabling a reasonable structural model to be inferred.

Community challenges in biomedical text mining over 10 years: success, failure and the future.

One effective way to improve the state of the art is through competitions. Following the success of the Critical Assessment of protein Structure Prediction (CASP) in bioinformatics research, a number of challenge evaluations have been organized by the text-mining research community to assess and advance natural language processing (NLP) research for biomedicine. In this article, we review the different community challenge evaluations held from 2002 to 2014 and their respective tasks. Furthermore, we examine these challenge tasks through their targeted problems in NLP research and biomedical applications, respectively. Next, we describe the general workflow of organizing a Biomedical NLP (BioNLP) challenge and involved stakeholders (task organizers, task data producers, task participants and end users). Finally, we summarize the impact and contributions by taking into account different BioNLP challenges as a whole, followed by a discussion of their limitations and difficulties. We conclude with future trends in BioNLP challenge evaluations.

Improving accuracy of protein contact prediction using balanced network deconvolution.

Residue contact map is essential for protein three-dimensional structure determination. But most of the current contact prediction methods based on residue co-evolution suffer from high false-positives as introduced by indirect and transitive contacts (i.e., residues A-B and B-C are in contact, but A-C are not). Built on the work by Feizi et al. (Nat Biotechnol 2013; 31:726-733), which demonstrated a general network model to distinguish direct dependencies by network deconvolution, this study presents a new balanced network deconvolution (BND) algorithm to identify optimized dependency matrix without limit on the eigenvalue range in the applied network systems. The algorithm was used to filter contact predictions of five widely used co-evolution methods. On the test of proteins from three benchmark datasets of the 9th critical assessment of protein structure prediction (CASP9), CASP10, and PSICOV (precise structural contact prediction using sparse inverse covariance estimation) database experiments, the BND can improve the medium- and long-range contact predictions at the L/5 cutoff by 55.59% and 47.68%, respectively, without additional central processing unit cost. The improvement is statistically significant, with a P-value < 5.93 × 10(-3) in the Student's t-test. A further comparison with the ab initio structure predictions in CASPs showed that the usefulness of the current co-evolution-based contact prediction to the three-dimensional structure modeling relies on the number of homologous sequences existing in the sequence databases. BND can be used as a general contact refinement method, which is freely available at: http://www.csbio.sjtu.edu.cn/bioinf/BND/.