Improving protein tertiary structure prediction by deep learning and distance prediction in CASP14

Substantial progresses in protein structure prediction have been made by utilizing deep-learning and residue-residue distance prediction since CASP13. Inspired by the advances, we improve our CASP14 MULTICOM protein structure prediction system by incorporating three new components: (a) a new deep learning-based protein inter-residue distance predictor to improve template-free (ab initio) tertiary structure prediction, (b) an enhanced template-based tertiary structure prediction method, and (c) distance-based model quality assessment methods empowered by deep learning. In the 2020 CASP14 experiment, MULTICOM predictor was ranked seventh out of 146 predictors in tertiary structure prediction and ranked third out of 136 predictors in inter-domain structure prediction. The results demonstrate that the template-free modeling based on deep learning and residue-residue distance prediction can predict the correct topology for almost all template-based modeling targets and a majority of hard targets (template-free targets or targets whose templates cannot be recognized), which is a significant improvement over the CASP13 MULTICOM predictor. Moreover, the template-free modeling performs better than the template-based modeling on not only hard targets but also the targets that have homologous templates. The performance of the template-free modeling largely depends on the accuracy of distance prediction closely related to the quality of multiple sequence alignments. The structural model quality assessment works well on targets for which enough good models can be predicted, but it may perform poorly when only a few good models are predicted for a hard target and the distribution of model quality scores is highly skewed. MULTICOM is available at https://github.com/jianlin-cheng/MULTICOM_Human_CASP14/tree/CASP14_DeepRank3 and https://github.com/multicom-toolbox/multicom/tree/multicom_v2.0 .     

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.     

Modeling large regions in proteins: applications to loops, termini, and folding

Template-based methods for predicting protein structure provide models for a significant portion of the protein but often contain insertions or chain ends (InsEnds) of indeterminate conformation. The local structure prediction "problem" entails modeling the InsEnds onto the rest of the protein. A well-known limit involves predicting loops of ≤12 residues in crystal structures. However, InsEnds may contain as many as ~50 amino acids, and the template-based model of the protein itself may be imperfect. To address these challenges, we present a free modeling method for predicting the local structure of loops and large InsEnds in both crystal structures and template-based models. The approach uses single amino acid torsional angle "pivot" moves of the protein backbone with a C(β) level representation. Nevertheless, our accuracy for loops is comparable to existing methods. We also apply a more stringent test, the blind structure prediction and refinement categories of the CASP9 tournament, where we improve the quality of several homology based models by modeling InsEnds as long as 45 amino acids, sizes generally inaccessible to existing loop prediction methods. Our approach ranks as one of the best in the CASP9 refinement category that involves improving template-based models so that they can function as molecular replacement models to solve the phase problem for crystallographic structure determination.     

Toward the solution of the protein structure prediction problem

Since Anfinsen demonstrated that the information encoded in a protein’s amino acid sequence determines its structure in 1973, solving the protein structure prediction problem has been the Holy Grail of structural biology. The goal of protein structure prediction approaches is to utilize computational modeling to determine the spatial location of every atom in a protein molecule starting from only its amino acid sequence. Depending on whether homologous structures can be found in the Protein Data Bank (PDB), structure prediction methods have been historically categorized as template-based modeling (TBM) or template-free modeling (FM) approaches. Until recently, TBM has been the most reliable approach to predicting protein structures, and in the absence of reliable templates, the modeling accuracy sharply declines. Nevertheless, the results of the most recent community-wide assessment of protein structure prediction experiment (CASP14) have demonstrated that the protein structure prediction problem can be largely solved through the use of end-to-end deep machine learning techniques, where correct folds could be built for nearly all single-domain proteins without using the PDB templates. Critically, the model quality exhibited little correlation with the quality of available template structures, as well as the number of sequence homologs detected for a given target protein. Thus, the implementation of deep-learning techniques has essentially broken through the 50-year-old modeling border between TBM and FM approaches and has made the success of high-resolution structure prediction significantly less dependent on template availability in the PDB library.     

基于模板的蛋白质结构预测

基于模板的蛋白结构预测和不依赖模板的蛋白结构预测是计算预测蛋白质三维结构的两种方法,前者由于具有快速和较高准确性的优点,而得到了广泛的应用.基于模板的结构预测是通过寻找与目标蛋白序列相似并且有实验测定的结构作为模板,进而构建目标序列的结构模型的方法.文章详细综述了基于模板的结构预测方法的步骤、关键环节,并对影响结构预测...     

Modularity of Protein Folds as a Tool for Template-Free Modeling of Structures

Predicting the three-dimensional structure of proteins from their amino acid sequences remains a challenging problem in molecular biology. While the current structural coverage of proteins is almost exclusively provided by template-based techniques, the modeling of the rest of the protein sequences increasingly require template-free methods. However, template-free modeling methods are much less reliable and are usually applicable for smaller proteins, leaving much space for improvement. We present here a novel computational method that uses a library of supersecondary structure fragments, known as Smotifs, to model protein structures. The library of Smotifs has saturated over time, providing a theoretical foundation for efficient modeling. The method relies on weak sequence signals from remotely related protein structures to create a library of Smotif fragments specific to the target protein sequence. This Smotif library is exploited in a fragment assembly protocol to sample decoys, which are assessed by a composite scoring function. Since the Smotif fragments are larger in size compared to the ones used in other fragment-based methods, the proposed modeling algorithm, SmotifTF, can employ an exhaustive sampling during decoy assembly. SmotifTF successfully predicts the overall fold of the target proteins in about 50% of the test cases and performs competitively when compared to other state of the art prediction methods, especially when sequence signal to remote homologs is diminishing. Smotif-based modeling is complementary to current prediction methods and provides a promising direction in addressing the structure prediction problem, especially when targeting larger proteins for modeling.     

Structural basis of GC-1 selectivity for thyroid hormone receptor isoforms

Background

Multiple protein templates are commonly used in manual protein structure prediction. However, few automated algorithms of selecting and combining multiple templates are available.

Results

Here we develop an effective multi-template combination algorithm for protein comparative modeling. The algorithm selects templates according to the similarity significance of the alignments between template and target proteins. It combines the whole template-target alignments whose similarity significance score is close to that of the top template-target alignment within a threshold, whereas it only takes alignment fragments from a less similar template-target alignment that align with a sizable uncovered region of the target. We compare the algorithm with the traditional method of using a single top template on the 45 comparative modeling targets (i.e. easy template-based modeling targets) used in the seventh edition of Critical Assessment of Techniques for Protein Structure Prediction (CASP7). The multi-template combination algorithm improves the GDT-TS scores of predicted models by 6.8% on average. The statistical analysis shows that the improvement is significant (p-value < 10^-4). Compared with the ideal approach that always uses the best template, the multi-template approach yields only slightly better performance. During the CASP7 experiment, the preliminary implementation of the multi-template combination algorithm (FOLDpro) was ranked second among 67 servers in the category of high-accuracy structure prediction in terms of GDT-TS measure.

Conclusion

We have developed a novel multi-template algorithm to improve protein comparative modeling.     

RNA and protein 3D structure modeling: similarities and differences

In analogy to proteins, the function of RNA depends on its structure and dynamics, which are encoded in the linear sequence. While there are numerous methods for computational prediction of protein 3D structure from sequence, there have been very few such methods for RNA. This review discusses template-based and template-free approaches for macromolecular structure prediction, with special emphasis on comparison between the already tried-and-tested methods for protein structure modeling and the very recently developed “protein-like” modeling methods for RNA. We highlight analogies between many successful methods for modeling of these two types of biological macromolecules and argue that RNA 3D structure can be modeled using “protein-like” methodology. We also highlight the areas where the differences between RNA and proteins require the development of RNA-specific solutions.     

Cover Image,Volume 87, Issue 7

Template-based modeling is considered as one of the most successful approaches for protein structure prediction. However, reliably and accurately selecting optimal template proteins from a library of known protein structures having similar folds as the target protein and making correct alignments between the target sequence and the template structures, a template-based modeling technique known as threading, remains challenging, particularly for non- or distantly-homologous protein targets. With the recent advancement in protein residue-residue contact map prediction powered by sequence co-evolution and machine learning, here we systematically analyze the effect of inclusion of residue-residue contact information in improving the accuracy and reliability of protein threading. We develop a new threading algorithm by incorporating various sequential and structural features, and subsequently integrate residue-residue contact information as an additional scoring term for threading template selection. We show that the inclusion of contact information attains statistically significantly better threading performance compared to a baseline threading algorithm that does not utilize contact information when everything else remains the same. Experimental results demonstrate that our contact based threading approach outperforms popular threading method MUSTER, contact-assisted ab initio folding method CONFOLD2, and recent state-of-the-art contact-assisted protein threading methods EigenTHREADER and map_align on several benchmarks. Our study illustrates that the inclusion of contact maps is a promising avenue in protein threading to ultimately help to improve the accuracy of protein structure prediction.     

(PS)2-v2: template-based protein structure prediction server

Background  

Template selection and target-template alignment are critical steps for template-based modeling (TBM) methods. To identify the template for the twilight zone of 15~25% sequence similarity between targets and templates is still difficulty for template-based protein structure prediction. This study presents the (PS)²-v2 server, based on our original server with numerous enhancements and modifications, to improve reliability and applicability.     

Homology-based modeling of 3D structures of protein-protein complexes using alignments of modified sequence profiles

Customary practice in predicting 3D structures of protein-protein complexes is employment of various docking methods when the structures of separate monomers are known a priori. The alternative approach, i.e. the template-based prediction with pure sequence information as a starting point, is still considered as being inferior mostly due to presumption that the pool of available structures of protein-protein complexes, which can serve as putative templates, is not sufficiently large. Recently, however, several labs have developed databases containing thousands of 3D structures of protein-protein complexes, which enable statistically reliable testing of homology-based algorithms. In this paper we report the results on homology-based modeling of 3D structures of protein complexes using alignments of modified sequence profiles. The method, called HOMology-BAsed COmplex Prediction (HOMBACOP), has two distinctive features: (I) extra weight on aligning interfacial residues in the dynamical programming algorithm, and (II) increased gap penalties for the interfacial segments. The method was tested against our recently developed ProtCom database and against the Boston University protein-protein BENCHMARK. In both cases, models generated were compared to the models built on basis of customarily protein structure initiative (PSI)-BLAST sequence alignments. It was found that existence of homologous (by the means of PSI-BLAST) templates (44% of cases) enables both methods to produce models of good quality, with the profiles method outperforming the PSI-BLAST models (with respect to the percentage of correctly predicted residues on the complex interface and fraction of native interfacial contacts). The models were evaluated according to the CAPRI assessment criteria and about two thirds of the models were found to fall into acceptable and medium-quality categories. The same comparison of a larger set of 463 protein complexes showed again that profiles generate better models. We further demonstrate, using our ProtCom database, the suitability of the profile alignment algorithm in detecting remote homologues between query and template sequences, where the PSI-BLAST method fails.     

Progress and challenges in protein structure prediction

Depending on whether similar structures are found in the PDB library, the protein structure prediction can be categorized into template-based modeling and free modeling. Although threading is an efficient tool to detect the structural analogs, the advancements in methodology development have come to a steady state. Encouraging progress is observed in structure refinement which aims at drawing template structures closer to the native; this has been mainly driven by the use of multiple structure templates and the development of hybrid knowledge-based and physics-based force fields. For free modeling, exciting examples have been witnessed in folding small proteins to atomic resolutions. However, predicting structures for proteins larger than 150 residues still remains a challenge, with bottlenecks from both force field and conformational search.     

Refinement of unreliable local regions in template-based protein models

Contemporary template-based modeling techniques allow applications of modeling methods to vast biological problems. However, they tend to fail to provide accurate structures for less-conserved local regions in sequence even when the overall structure can be modeled reliably. We call these regions unreliable local regions (ULRs). Accurate modeling of ULRs is of enormous value because they are frequently involved in functional specificity. In this article, we introduce a new method for modeling ULRs in template-based models by employing a sophisticated loop modeling technique. Combined with our previous study on protein termini, the method is applicable to refinement of both loop and terminus ULRs. A large-scale test carried out in a blind fashion in CASP9 (the 9th Critical Assessment of techniques for protein structure prediction) shows that ULR structures are improved over initial template-based models by refinement in more than 70% of the successfully detected ULRs. It is also notable that successful modeling of several long ULRs over 12 residues is achieved. Overall, the current results show that a careful application of loop and terminus modeling can be a promising tool for model refinement in template-based modeling.     

A decade of CASP: progress, bottlenecks and prognosis in protein structure prediction

For the past ten years, CASP (Critical Assessment of Structure Prediction) has monitored the state of the art in modeling protein structure from sequence. During this period, there has been substantial progress in both comparative modeling of structure (using information from an evolutionarily related structural template) and template-free modeling. The quality of comparative models depends on the closeness of the evolutionary relationship on which they are based. Template-free modeling, although still very approximate, now produces topologically near correct models for some small proteins. Current major challenges are refining comparative models so that they match experimental accuracy, obtaining accurate sequence alignments for models based on remote evolutionary relationships, and extending template-free modeling methods so that they produce more accurate models, handle parts of comparative models not available from a template and deal with larger structures.     

Deep template-based protein structure prediction

MotivationProtein structure prediction has been greatly improved by deep learning, but most efforts are devoted to template-free modeling. But very few deep learning methods are developed for TBM (template-based modeling), a popular technique for protein structure prediction. TBM has been studied extensively in the past, but its accuracy is not satisfactory when highly similar templates are not available.ResultsThis paper presents a new method NDThreader (New Deep-learning Threader) to address the challenges of TBM. NDThreader first employs DRNF (deep convolutional residual neural fields), which is an integration of deep ResNet (convolutional residue neural networks) and CRF (conditional random fields), to align a query protein to templates without using any distance information. Then NDThreader uses ADMM (alternating direction method of multipliers) and DRNF to further improve sequence-template alignments by making use of predicted distance potential. Finally, NDThreader builds 3D models from a sequence-template alignment by feeding it and sequence coevolution information into a deep ResNet to predict inter-atom distance distribution, which is then fed into PyRosetta for 3D model construction. Our experimental results show that NDThreader greatly outperforms existing methods such as CNFpred, HHpred, DeepThreader and CEthreader. NDThreader was blindly tested in CASP14 as a part of RaptorX server, which obtained the best average GDT score among all CASP14 servers on the 58 TBM targets.     

eThread: A Highly Optimized Machine Learning-Based Approach to Meta-Threading and the Modeling of Protein Tertiary Structures

Template-based modeling that employs various meta-threading techniques is currently the most accurate, and consequently the most commonly used, approach for protein structure prediction. Despite the evident progress in this field, accurate structure models cannot be constructed for a significant fraction of gene products, thus the development of new algorithms is required. Here, we describe the development, optimization and large-scale benchmarking of eThread, a highly accurate meta-threading procedure for the identification of structural templates and the construction of corresponding target-to-template alignments. eThread integrates ten state-of-the-art threading/fold recognition algorithms in a local environment and extensively uses various machine learning techniques to carry out fully automated template-based protein structure modeling. Tertiary structure prediction employs two protocols based on widely used modeling algorithms: Modeller and TASSER-Lite. As a part of eThread, we also developed eContact, which is a Bayesian classifier for the prediction of inter-residue contacts and eRank, which effectively ranks generated multiple protein models and provides reliable confidence estimates as structure quality assessment. Excluding closely related templates from the modeling process, eThread generates models, which are correct at the fold level, for >80% of the targets; 40–50% of the constructed models are of a very high quality, which would be considered accurate at the family level. Furthermore, in large-scale benchmarking, we compare the performance of eThread to several alternative methods commonly used in protein structure prediction. Finally, we estimate the upper bound for this type of approach and discuss the directions towards further improvements.     

Improving protein template recognition by using small-angle x-ray scattering profiles

Small-angle x-ray scattering (SAXS) is able to extract low-resolution protein shape information without requiring a specific crystal formation. However, it has found little use in atomic-level protein structure determination due to the uncertainty of residue-level structural assignment. We developed a new algorithm, SAXSTER, to couple the raw SAXS data with protein-fold-recognition algorithms and thus improve template-based protein-structure predictions. We designed nine different matching scoring functions of template and experimental SAXS profiles. The logarithm of the integrated correlation score showed the best template recognition ability and had the highest correlation with the true template modeling (TM)-score of the target structures. We tested the method in large-scale protein-fold-recognition experiments and achieved significant improvements in prioritizing the best template structures. When SAXSTER was applied to the proteins of asymmetric SAXS profile distributions, the average TM-score of the top-ranking templates increased by 18% after homologous templates were excluded, which corresponds to a p-value < 10⁻⁹ in Student's t-test. These data demonstrate a promising use of SAXS data to facilitate computational protein structure modeling, which is expected to work most efficiently for proteins of irregular global shape and/or multiple-domain protein complexes.     

GWYRE: A Resource for Mapping Variants onto Experimental and Modeled Structures of Human Protein Complexes

Rapid progress in structural modeling of proteins and their interactions is powered by advances in knowledge-based methodologies along with better understanding of physical principles of protein structure and function. The pool of structural data for modeling of proteins and protein–protein complexes is constantly increasing due to the rapid growth of protein interaction databases and Protein Data Bank. The GWYRE (Genome Wide PhYRE) project capitalizes on these developments by advancing and applying new powerful modeling methodologies to structural modeling of protein–protein interactions and genetic variation. The methods integrate knowledge-based tertiary structure prediction using Phyre2 and quaternary structure prediction using template-based docking by a full-structure alignment protocol to generate models for binary complexes. The predictions are incorporated in a comprehensive public resource for structural characterization of the human interactome and the location of human genetic variants. The GWYRE resource facilitates better understanding of principles of protein interaction and structure/function relationships. The resource is available at http://www.gwyre.org.     

A multiple-template approach to protein threading

Most threading methods predict the structure of a protein using only a single template. Due to the increasing number of solved structures, a protein without solved structure is very likely to have more than one similar template structures. Therefore, a natural question to ask is if we can improve modeling accuracy using multiple templates. This article describes a new multiple-template threading method to answer this question. At the heart of this multiple-template threading method is a novel probabilistic-consistency algorithm that can accurately align a single protein sequence simultaneously to multiple templates. Experimental results indicate that our multiple-template method can improve pairwise sequence-template alignment accuracy and generate models with better quality than single-template models even if they are built from the best single templates (P-value <10(-6)) while many popular multiple sequence/structure alignment tools fail to do so. The underlying reason is that our probabilistic-consistency algorithm can generate accurate multiple sequence/template alignments. In another word, without an accurate multiple sequence/template alignment, the modeling accuracy cannot be improved by simply using multiple templates to increase alignment coverage. Blindly tested on the CASP9 targets with more than one good template structures, our method outperforms all other CASP9 servers except two (Zhang-Server and QUARK of the same group). Our probabilistic-consistency algorithm can possibly be extended to align multiple protein/RNA sequences and structures.     

Multibody coarse-grained potentials for native structure recognition and quality assessment of protein models

Multibody potentials have been of much interest recently because they take into account three dimensional interactions related to residue packing and capture the cooperativity of these interactions in protein structures. Our goal was to combine long range multibody potentials and short range potentials to improve recognition of native structure among misfolded decoys. We optimized the weights for four-body nonsequential, four-body sequential, and short range potentials to obtain optimal model ranking results for threading and have compared these data against results obtained with other potentials (26 different coarse-grained potentials from the Potentials 'R'Us web server have been used). Our optimized multibody potentials outperform all other contact potentials in the recognition of the native structure among decoys, both for models from homology template-based modeling and from template-free modeling in CASP8 decoy sets. We have compared the results obtained for this optimized coarse-grained potentials, where each residue is represented by a single point, with results obtained by using the DFIRE potential, which takes into account atomic level information of proteins. We found that for all proteins larger than 80 amino acids our optimized coarse-grained potentials yield results comparable to those obtained with the atomic DFIRE potential.