Princeton_TIGRESS: Protein geometry refinement using simulations and support vector machines

Protein structure refinement aims to perform a set of operations given a predicted structure to improve model quality and accuracy with respect to the native in a blind fashion. Despite the numerous computational approaches to the protein refinement problem reported in the previous three CASPs, an overwhelming majority of methods degrade models rather than improve them. We initially developed a method tested using blind predictions during CASP10 which was officially ranked in 5th place among all methods in the refinement category. Here, we present Princeton_TIGRESS, which when benchmarked on all CASP 7,8,9, and 10 refinement targets, simultaneously increased GDT_TS 76% of the time with an average improvement of 0.83 GDT_TS points per structure. The method was additionally benchmarked on models produced by top performing three‐dimensional structure prediction servers during CASP10. The robustness of the Princeton_TIGRESS protocol was also tested for different random seeds. We make the Princeton_TIGRESS refinement protocol freely available as a web server at http://atlas.princeton.edu/refinement . Using this protocol, one can consistently refine a prediction to help bridge the gap between a predicted structure and the actual native structure. Proteins 2014; 82:794–814. © 2013 Wiley Periodicals, Inc.     

Analysis of distance-based protein structure prediction by deep learning in CASP13

This paper reports the CASP13 results of distance-based contact prediction, threading, and folding methods implemented in three RaptorX servers, which are built upon the powerful deep convolutional residual neural network (ResNet) method initiated by us for contact prediction in CASP12. On the 32 CASP13 FM (free-modeling) targets with a median multiple sequence alignment (MSA) depth of 36, RaptorX yielded the best contact prediction among 46 groups and almost the best 3D structure modeling among all server groups without time-consuming conformation sampling. In particular, RaptorX achieved top L/5, L/2, and L long-range contact precision of 70%, 58%, and 45%, respectively, and predicted correct folds (TMscore > 0.5) for 18 of 32 targets. Further, RaptorX predicted correct folds for all FM targets with >300 residues (T0950-D1, T0969-D1, and T1000-D2) and generated the best 3D models for T0950-D1 and T0969-D1 among all groups. This CASP13 test confirms our previous findings: (a) predicted distance is more useful than contacts for both template-based and free modeling; and (b) structure modeling may be improved by integrating template and coevolutionary information via deep learning. This paper will discuss progress we have made since CASP12, the strength and weakness of our methods, and why deep learning performed much better in CASP13.     

Protein tertiary structure modeling driven by deep learning and contact distance prediction in CASP13

Predicting residue-residue distance relationships (eg, contacts) has become the key direction to advance protein structure prediction since 2014 CASP11 experiment, while deep learning has revolutionized the technology for contact and distance distribution prediction since its debut in 2012 CASP10 experiment. During 2018 CASP13 experiment, we enhanced our MULTICOM protein structure prediction system with three major components: contact distance prediction based on deep convolutional neural networks, distance-driven template-free (ab initio) modeling, and protein model ranking empowered by deep learning and contact prediction. Our experiment demonstrates that contact distance prediction and deep learning methods are the key reasons that MULTICOM was ranked 3rd out of all 98 predictors in both template-free and template-based structure modeling in CASP13. Deep convolutional neural network can utilize global information in pairwise residue-residue features such as coevolution scores to substantially improve contact distance prediction, which played a decisive role in correctly folding some free modeling and hard template-based modeling targets. Deep learning also successfully integrated one-dimensional structural features, two-dimensional contact information, and three-dimensional structural quality scores to improve protein model quality assessment, where the contact prediction was demonstrated to consistently enhance ranking of protein models for the first time. The success of MULTICOM system clearly shows that protein contact distance prediction and model selection driven by deep learning holds the key of solving protein structure prediction problem. However, there are still challenges in accurately predicting protein contact distance when there are few homologous sequences, folding proteins from noisy contact distances, and ranking models of hard targets.     

CASP13 target classification into tertiary structure prediction categories

Protein target structures for the Critical Assessment of Structure Prediction round 13 (CASP13) were split into evaluation units (EUs) based on their structural domains, the domain organization of available templates, and the performance of servers on whole targets compared to split target domains. Eighty targets were split into 112 EUs. The EUs were classified into categories suitable for assessment of high accuracy modeling (or template-based modeling [TBM]) and topology (or free modeling [FM]) based on target difficulty. Assignment into assessment categories considered the following criteria: (a) the evolutionary relationship of target domains to existing fold space as defined by the Evolutionary Classification of Protein Domains (ECOD) database; (b) the clustering of target domains using eight objective sequence, structure, and performance measures; and (c) the placement of target domains in a scatter plot of target difficulty against server performance used in the previous CASP. Generally, target domains with good server predictions had close template homologs and were classified as TBM. Alternately, targets with poor server predictions represent a mixture of fast evolving homologs, structure analogs, and new folds, and were classified as FM or FM/TBM overlap.     

Protein structure prediction assisted with sparse NMR data in CASP13

CASP13 has investigated the impact of sparse NMR data on the accuracy of protein structure prediction. NOESY and ¹⁵N-¹H residual dipolar coupling data, typical of that obtained for ¹⁵N,¹³C-enriched, perdeuterated proteins up to about 40 kDa, were simulated for 11 CASP13 targets ranging in size from 80 to 326 residues. For several targets, two prediction groups generated models that are more accurate than those produced using baseline methods. Real NMR data collected for a de novo designed protein were also provided to predictors, including one data set in which only backbone resonance assignments were available. Some NMR-assisted prediction groups also did very well with these data. CASP13 also assessed whether incorporation of sparse NMR data improves the accuracy of protein structure prediction relative to nonassisted regular methods. In most cases, incorporation of sparse, noisy NMR data results in models with higher accuracy. The best NMR-assisted models were also compared with the best regular predictions of any CASP13 group for the same target. For six of 13 targets, the most accurate model provided by any NMR-assisted prediction group was more accurate than the most accurate model provided by any regular prediction group; however, for the remaining seven targets, one or more regular prediction method provided a more accurate model than even the best NMR-assisted model. These results suggest a novel approach for protein structure determination, in which advanced prediction methods are first used to generate structural models, and sparse NMR data is then used to validate and/or refine these models.     

Consistent refinement of submitted models at CASP using a knowledge‐based potential

Protein structure refinement is an important but unsolved problem; it must be solved if we are to predict biological function that is very sensitive to structural details. Specifically, critical assessment of techniques for protein structure prediction (CASP) shows that the accuracy of predictions in the comparative modeling category is often worse than that of the template on which the homology model is based. Here we describe a refinement protocol that is able to consistently refine submitted predictions for all categories at CASP7. The protocol uses direct energy minimization of the knowledge‐based potential of mean force that is based on the interaction statistics of 167 atom types (Summa and Levitt, Proc Natl Acad Sci USA 2007; 104:3177–3182). Our protocol is thus computationally very efficient; it only takes a few minutes of CPU time to run typical protein models (300 residues). We observe an average structural improvement of 1% in GDT_TS, for predictions that have low and medium homology to known PDB structures (Global Distance Test score or GDT_TS between 50 and 80%). We also observe a marked improvement in the stereochemistry of the models. The level of improvement varies amongst the various participants at CASP, but we see large improvements (>10% increase in GDT_TS) even for models predicted by the best performing groups at CASP7. In addition, our protocol consistently improved the best predicted models in the refinement category at CASP7 and CASP8. These improvements in structure and stereochemistry prove the usefulness of our computationally inexpensive, powerful and automatic refinement protocol. Proteins 2010. © 2010 Wiley‐Liss, Inc.     

FunFOLDQA: a quality assessment tool for protein-ligand binding site residue predictions

The estimation of prediction quality is important because without quality measures, it is difficult to determine the usefulness of a prediction. Currently, methods for ligand binding site residue predictions are assessed in the function prediction category of the biennial Critical Assessment of Techniques for Protein Structure Prediction (CASP) experiment, utilizing the Matthews Correlation Coefficient (MCC) and Binding-site Distance Test (BDT) metrics. However, the assessment of ligand binding site predictions using such metrics requires the availability of solved structures with bound ligands. Thus, we have developed a ligand binding site quality assessment tool, FunFOLDQA, which utilizes protein feature analysis to predict ligand binding site quality prior to the experimental solution of the protein structures and their ligand interactions. The FunFOLDQA feature scores were combined using: simple linear combinations, multiple linear regression and a neural network. The neural network produced significantly better results for correlations to both the MCC and BDT scores, according to Kendall's τ, Spearman's ρ and Pearson's r correlation coefficients, when tested on both the CASP8 and CASP9 datasets. The neural network also produced the largest Area Under the Curve score (AUC) when Receiver Operator Characteristic (ROC) analysis was undertaken for the CASP8 dataset. Furthermore, the FunFOLDQA algorithm incorporating the neural network, is shown to add value to FunFOLD, when both methods are employed in combination. This results in a statistically significant improvement over all of the best server methods, the FunFOLD method (6.43%), and one of the top manual groups (FN293) tested on the CASP8 dataset. The FunFOLDQA method was also found to be competitive with the top server methods when tested on the CASP9 dataset. To the best of our knowledge, FunFOLDQA is the first attempt to develop a method that can be used to assess ligand binding site prediction quality, in the absence of experimental data.     

A further leap of improvement in tertiary structure prediction in CASP13 prompts new routes for future assessments

We present our assessment of tertiary structure predictions for hard targets in Critical Assessment of Structure Prediction round 13 (CASP13). The analysis includes (a) assignment and discussion of best models through scores-aided visual inspection of models for each evaluation unit (EU); (b) ranking of predictors resulting from this evaluation and from global scores; and (c) evaluation of progress, state of the art, and current limitations of protein structure prediction. We witness a sizable improvement in tertiary structure prediction building on the progress observed from CASP11 to CASP12, with (a) top models reaching backbone RMSD <3 å for several EUs of size <150 residues, contributed by many groups; (b) at least one model that roughly captures global topology for all EUs, probably unprecedented in this track of CASP; and (c) even quite good models for full, unsplit targets. Better structure predictions are brought about mainly by improved residue-residue contact predictions, and since this CASP also by distance predictions, achieved through state-of-the-art machine learning methods which also progressed to work with slightly shallower alignments compared to CASP12. As we reach a new realm of tertiary structure prediction quality, new directions are proposed and explored for future CASPs: (a) dropping splitting into EUs, (b) rethinking difficulty metrics probably in terms of contact and distance predictions, (c) assessing also side chains for models of high backbone accuracy, and (d) assessing residue-wise and possibly residue-residue quality estimates.     

Evaluation of template-based modeling in CASP13

Performance in the template-based modeling (TBM) category of CASP13 is assessed here, using a variety of metrics. Performance of the predictor groups that participated is ranked using the primary ranking score that was developed by the assessors for CASP12. This reveals that the best results are obtained by groups that include contact predictions or inter-residue distance predictions derived from deep multiple sequence alignments. In cases where there is a good homolog in the wwPDB (TBM-easy category), the best results are obtained by modifying a template. However, for cases with poorer homologs (TBM-hard), very good results can be obtained without using an explicit template, by deep learning algorithms trained on the wwPDB. Alternative metrics are introduced, to allow testing of aspects of structural models that are not addressed by traditional CASP metrics. These include comparisons to the main-chain and side-chain torsion angles of the target, and the utility of models for solving crystal structures by the molecular replacement method. The alternative metrics are poorly correlated with the traditional metrics, and it is proposed that modeling has reached a sufficient level of maturity that the best models should be expected to satisfy this wider range of criteria.     

Deep-learning contact-map guided protein structure prediction in CASP13

We report the results of two fully automated structure prediction pipelines, “Zhang-Server” and “QUARK”, in CASP13. The pipelines were built upon the C-I-TASSER and C-QUARK programs, which in turn are based on I-TASSER and QUARK but with three new modules: (a) a novel multiple sequence alignment (MSA) generation protocol to construct deep sequence-profiles for contact prediction; (b) an improved meta-method, NeBcon, which combines multiple contact predictors, including ResPRE that predicts contact-maps by coupling precision-matrices with deep residual convolutional neural-networks; and (c) an optimized contact potential to guide structure assembly simulations. For 50 CASP13 FM domains that lacked homologous templates, average TM-scores of the first models produced by C-I-TASSER and C-QUARK were 28% and 56% higher than those constructed by I-TASSER and QUARK, respectively. For the first time, contact-map predictions demonstrated usefulness on TBM domains with close homologous templates, where TM-scores of C-I-TASSER models were significantly higher than those of I-TASSER models with a P-value <.05. Detailed data analyses showed that the success of C-I-TASSER and C-QUARK was mainly due to the increased accuracy of deep-learning-based contact-maps, as well as the careful balance between sequence-based contact restraints, threading templates, and generic knowledge-based potentials. Nevertheless, challenges still remain for predicting quaternary structure of multi-domain proteins, due to the difficulties in domain partitioning and domain reassembly. In addition, contact prediction in terminal regions was often unsatisfactory due to the sparsity of MSAs. Development of new contact-based domain partitioning and assembly methods and training contact models on sparse MSAs may help address these issues.     

Assessing the accuracy of contact predictions in CASP13

The accuracy of sequence-based tertiary contact predictions was assessed in a blind prediction experiment at the CASP13 meeting. After 4 years of significant improvements in prediction accuracy, another dramatic advance has taken place since CASP12 was held 2 years ago. The precision of predicting the top L/5 contacts in the free modeling category, where L is the corresponding length of the protein in residues, has exceeded 70%. As a comparison, the best-performing group at CASP12 with a 47% precision would have finished below the top 1/3 of the CASP13 groups. Extensively trained deep neural network approaches dominate the top performing algorithms, which appear to efficiently integrate information on coevolving residues and interacting fragments or possibly utilize memories of sequence similarities and sometimes can deliver accurate results even in the absence of virtually any target specific evolutionary information. If the current performance is evaluated by F-score on L contacts, it stands around 24% right now, which, despite the tremendous impact and advance in improving its utility for structure modeling, also suggests that there is much room left for further improvement.     

Prediction of interresidue contacts with DeepMetaPSICOV in CASP13

In this article, we describe our efforts in contact prediction in the CASP13 experiment. We employed a new deep learning-based contact prediction tool, DeepMetaPSICOV (or DMP for short), together with new methods and data sources for alignment generation. DMP evolved from MetaPSICOV and DeepCov and combines the input feature sets used by these methods as input to a deep, fully convolutional residual neural network. We also improved our method for multiple sequence alignment generation and included metagenomic sequences in the search. We discuss successes and failures of our approach and identify areas where further improvements may be possible. DMP is freely available at: https://github.com/psipred/DeepMetaPSICOV .     

Assessment of protein model structure accuracy estimation in CASP13: Challenges in the era of deep learning

Scoring model structure is an essential component of protein structure prediction that can affect the prediction accuracy tremendously. Users of protein structure prediction results also need to score models to select the best models for their application studies. In Critical Assessment of techniques for protein Structure Prediction (CASP), model accuracy estimation methods have been tested in a blind fashion by providing models submitted by the tertiary structure prediction servers for scoring. In CASP13, model accuracy estimation results were evaluated in terms of both global and local structure accuracy. Global structure accuracy estimation was evaluated by the quality of the models selected by the global structure scores and by the absolute estimates of the global scores. Residue-wise, local structure accuracy estimations were evaluated by three different measures. A new measure introduced in CASP13 evaluates the ability to predict inaccurately modeled regions that may be improved by refinement. An intensive comparative analysis on CASP13 and the previous CASPs revealed that the tertiary structure models generated by the CASP13 servers show very distinct features. Higher consensus toward models of higher global accuracy appeared even for free modeling targets, and many models of high global accuracy were not well optimized at the atomic level. This is related to the new technology in CASP13, deep learning for tertiary contact prediction. The tertiary model structures generated by deep learning pose a new challenge for EMA (estimation of model accuracy) method developers. Model accuracy estimation itself is also an area where deep learning can potentially have an impact, although current EMA methods have not fully explored that direction.     

Protein structural motif prediction in multidimensional phi-psi space leads to improved secondary structure prediction.

A significant step towards establishing the structure and function of a protein is the prediction of the local conformation of the polypeptide chain. In this article, we present systems for the prediction of three new alphabets of local structural motifs. The motifs are built by applying multidimensional scaling (MDS) and clustering to pair-wise angular distances for multiple phi-psi angle values collected from high-resolution protein structures. The predictive systems, based on ensembles of bidirectional recurrent neural network architectures, and trained on a large non-redundant set of protein structures, achieve 72%, 66%, and 60% correct motif prediction on an independent test set for di-peptides (six classes), tri-peptides (eight classes) and tetra-peptides (14 classes), respectively, 28-30% above baseline statistical predictors. We then build a further system, based on ensembles of two-layered bidirectional recurrent neural networks, to map structural motif predictions into a traditional 3-class (helix, strand, coil) secondary structure. This system achieves 79.5% correct prediction using the "hard" CASP 3-class assignment, and 81.4% with a more lenient assignment, outperforming a sophisticated state-of-the-art predictor (Porter) trained in the same experimental conditions. The structural motif predictor is publicly available at: http://distill.ucd.ie/porter+/.     

Prediction of protein secondary structure at 80% accuracy

Secondary structure prediction involving up to 800 neural network predictions has been developed, by use of novel methods such as output expansion and a unique balloting procedure. An overall performance of 77.2%-80.2% (77.9%-80.6% mean per-chain) for three-state (helix, strand, coil) prediction was obtained when evaluated on a commonly used set of 126 protein chains. The method uses profiles made by position-specific scoring matrices as input, while at the output level it predicts on three consecutive residues simultaneously. The predictions arise from tenfold, cross validated training and testing of 1032 protein sequences, using a scheme with primary structure neural networks followed by structure filtering neural networks. With respect to blind prediction, this work is preliminary and awaits evaluation by CASP4.     

Prediction of Protein Structure by Simulating Coarse-grained Folding Pathways: A Preliminary Report

Abstract

A set of software tools designed to study protein structure and kinetics has been developed. The core of these tools is a program called Folding Machine (FM) which is able to generate low resolution folding pathways using modest computational resources. The FM is based on a coarse-grained kinetic ab initio Monte-Carlo sampler that can optionally use information extracted from secondary structure prediction servers or from fragment libraries of local structure. The model underpinning this algorithm contains two novel elements: (a) the conformational space is discretized using the Ramachandran basins defined in the local φ-ψ energy maps; and (b) the solvent is treated implicitly by rescaling the pairwise terms of the non-bonded energy function according to the local solvent environments. The purpose of this hybrid ab initio/knowledge-based approach is threefold: to cover the long time scales of folding, to generate useful 3-dimensional models of protein structures, and to gain insight on the protein folding kinetics. Even though the algorithm is not yet fully developed, it has been used in a recent blind test of protein structure prediction (CASP5). The FM generated models within 6 Å backbone rmsd for fragments of about 60–70 residues of a-helical proteins. For a CASP5 target that turned out to be natively unfolded, the trajectory obtained for this sequence uniquely failed to converge. Also, a new measure to evaluate structure predictions is presented and used along the standard CASP assessment methods. Finally, recent improvements in the prediction of β-sheet structures are briefly described.     

Protein contact prediction using patterns of correlation

We describe a new method for using neural networks to predict residue contact pairs in a protein. The main inputs to the neural network are a set of 25 measures of correlated mutation between all pairs of residues in two "windows" of size 5 centered on the residues of interest. While the individual pair-wise correlations are a relatively weak predictor of contact, by training the network on windows of correlation the accuracy of prediction is significantly improved. The neural network is trained on a set of 100 proteins and then tested on a disjoint set of 1033 proteins of known structure. An average predictive accuracy of 21.7% is obtained taking the best L/2 predictions for each protein, where L is the sequence length. Taking the best L/10 predictions gives an average accuracy of 30.7%. The predictor is also tested on a set of 59 proteins from the CASP5 experiment. The accuracy is found to be relatively consistent across different sequence lengths, but to vary widely according to the secondary structure. Predictive accuracy is also found to improve by using multiple sequence alignments containing many sequences to calculate the correlations.     

CASP10–BCL::Fold efficiently samples topologies of large proteins

During CASP10 in summer 2012, we tested BCL::Fold for prediction of free modeling (FM) and template‐based modeling (TBM) targets. BCL::Fold assembles the tertiary structure of a protein from predicted secondary structure elements (SSEs) omitting more flexible loop regions early on. This approach enables the sampling of conformational space for larger proteins with more complex topologies. In preparation of CASP11, we analyzed the quality of CASP10 models throughout the prediction pipeline to understand BCL::Fold's ability to sample the native topology, identify native‐like models by scoring and/or clustering approaches, and our ability to add loop regions and side chains to initial SSE‐only models. The standout observation is that BCL::Fold sampled topologies with a GDT_TS score > 33% for 12 of 18 and with a topology score > 0.8 for 11 of 18 test cases de novo. Despite the sampling success of BCL::Fold, significant challenges still exist in clustering and loop generation stages of the pipeline. The clustering approach employed for model selection often failed to identify the most native‐like assembly of SSEs for further refinement and submission. It was also observed that for some β‐strand proteins model refinement failed as β‐strands were not properly aligned to form hydrogen bonds removing otherwise accurate models from the pool. Further, BCL::Fold samples frequently non‐natural topologies that require loop regions to pass through the center of the protein. Proteins 2015; 83:547–563. © 2015 Wiley Periodicals, Inc.     

Small angle X-ray scattering-assisted protein structure prediction in CASP13 and emergence of solution structure differences

Small angle X-ray scattering (SAXS) measures comprehensive distance information on a protein's structure, which can constrain and guide computational structure prediction algorithms. Here, we evaluate structure predictions of 11 monomeric and oligomeric proteins for which SAXS data were collected and provided to predictors in the 13th round of the Critical Assessment of protein Structure Prediction (CASP13). The category for SAXS-assisted predictions made gains in certain areas for CASP13 compared to CASP12. Improvements included higher quality data with size exclusion chromatography-SAXS (SEC-SAXS) and better selection of targets and communication of results by CASP organizers. In several cases, we can track improvements in model accuracy with use of SAXS data. For hard multimeric targets where regular folding algorithms were unsuccessful, SAXS data helped predictors to build models better resembling the global shape of the target. For most models, however, no significant improvement in model accuracy at the domain level was registered from use of SAXS data, when rigorously comparing SAXS-assisted models to the best regular server predictions. To promote future progress in this category, we identify successes, challenges, and opportunities for improved strategies in prediction, assessment, and communication of SAXS data to predictors. An important observation is that, for many targets, SAXS data were inconsistent with crystal structures, suggesting that these proteins adopt different conformation(s) in solution. This CASP13 result, if representative of PDB structures and future CASP targets, may have substantive implications for the structure training databases used for machine learning, CASP, and use of prediction models for biology.     

Deducing high-accuracy protein contact-maps from a triplet of coevolutionary matrices through deep residual convolutional networks

The topology of protein folds can be specified by the inter-residue contact-maps and accurate contact-map prediction can help ab initio structure folding. We developed TripletRes to deduce protein contact-maps from discretized distance profiles by end-to-end training of deep residual neural-networks. Compared to previous approaches, the major advantage of TripletRes is in its ability to learn and directly fuse a triplet of coevolutionary matrices extracted from the whole-genome and metagenome databases and therefore minimize the information loss during the course of contact model training. TripletRes was tested on a large set of 245 non-homologous proteins from CASP 11&12 and CAMEO experiments and outperformed other top methods from CASP12 by at least 58.4% for the CASP 11&12 targets and 44.4% for the CAMEO targets in the top-L long-range contact precision. On the 31 FM targets from the latest CASP13 challenge, TripletRes achieved the highest precision (71.6%) for the top-L/5 long-range contact predictions. It was also shown that a simple re-training of the TripletRes model with more proteins can lead to further improvement with precisions comparable to state-of-the-art methods developed after CASP13. These results demonstrate a novel efficient approach to extend the power of deep convolutional networks for high-accuracy medium- and long-range protein contact-map predictions starting from primary sequences, which are critical for constructing 3D structure of proteins that lack homologous templates in the PDB library.