An evaluation of automated homology modelling methods at low target template sequence similarity

MOTIVATION: There are two main areas of difficulty in homology modelling that are particularly important when sequence identity between target and template falls below 50%: sequence alignment and loop building. These problems become magnified with automatic modelling processes, as there is no human input to correct mistakes. As such we have benchmarked several stand-alone strategies that could be implemented in a workflow for automated high-throughput homology modelling. These include three new sequence-structure alignment programs: 3D-Coffee, Staccato and SAlign, plus five homology modelling programs and their respective loop building methods: Builder, Nest, Modeller, SegMod/ENCAD and Swiss-Model. The SABmark database provided 123 targets with at least five templates from the same SCOP family and sequence identities 相似文献   

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.     

Model quality assessment using distance constraints from alignments

Given a set of alternative models for a specific protein sequence, the model quality assessment (MQA) problem asks for an assignment of scores to each model in the set. A good MQA program assigns these scores such that they correlate well with real quality of the models, ideally scoring best that model which is closest to the true structure. In this article, we present a new approach for addressing the MQA problem. It is based on distance constraints extracted from alignments to templates of known structure, and is implemented in the Undertaker program for protein structure prediction. One novel feature is that we extract noncontact constraints as well as contact constraints. We describe how the distance constraint extraction is done and we show how they can be used to address the MQA problem. We have compared our method on CASP7 targets and the results show that our method is at least comparable with the best MQA methods that were assessed at CASP7. We also propose a new evaluation measure, Kendall's τ, that is more interpretable than conventional measures used for evaluating MQA methods (Pearson's r and Spearman's ρ). We show clear examples where Kendall's τ agrees much more with our intuition of a correct MQA, and we therefore propose that Kendall's τ be used for future CASP MQA assessments. Proteins 2009. © 2008 Wiley‐Liss, Inc.     

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 .     

A template-finding algorithm and a comprehensive benchmark for homology modeling of proteins

The first step in homology modeling is to identify a template protein for the target sequence. The template structure is used in later phases of the calculation to construct an atomically detailed model for the target. We have built from the Protein Data Bank (PDB) a large-scale learning set that includes tens of millions of pair matches that can be either a true template or a false one. Discriminatory learning (learning from positive and negative examples) is used to train a decision tree. Each branch of the tree is a mathematical programming model. The decision tree is tested on an independent set from PDB entries and on the sequences of CASP7. It provides significant enrichment of true templates (between 50 and 100%) when compared to PSI-BLAST. The model is further verified by building atomically detailed structures for each of the tentative true templates with modeller. The probability that a true match does not yield an acceptable structural model (within 6 A RMSD from the native structure) decays linearly as a function of the TM structural-alignment score.     

Effect of using suboptimal alignments in template-based protein structure prediction

Computational protein structure prediction remains a challenging task in protein bioinformatics. In the recent years, the importance of template-based structure prediction is increasing because of the growing number of protein structures solved by the structural genomics projects. To capitalize the significant efforts and investments paid on the structural genomics projects, it is urgent to establish effective ways to use the solved structures as templates by developing methods for exploiting remotely related proteins that cannot be simply identified by homology. In this work, we examine the effect of using suboptimal alignments in template-based protein structure prediction. We showed that suboptimal alignments are often more accurate than the optimal one, and such accurate suboptimal alignments can occur even at a very low rank of the alignment score. Suboptimal alignments contain a significant number of correct amino acid residue contacts. Moreover, suboptimal alignments can improve template-based models when used as input to Modeller. Finally, we use suboptimal alignments for handling a contact potential in a probabilistic way in a threading program, SUPRB. The probabilistic contacts strategy outperforms the partly thawed approach, which only uses the optimal alignment in defining residue contacts, and also the re-ranking strategy, which uses the contact potential in re-ranking alignments. The comparison with existing methods in the template-recognition test shows that SUPRB is very competitive and outperforms existing methods.     

Improved predictions by Pcons.net using multiple templates

Multiple templates can often be used to build more accurate homology models than models built from a single template. Here we introduce PconsM, an automated protocol that uses multiple templates to build protein models. PconsM has been among the top-performing methods in the recent CASP experiments and consistently perform better than the single template models used in Pcons.net. In particular for the easier targets with many alternative templates with a high degree of sequence identity, quality is readily improved with a few percentages over the highest ranked model built on a single template. PconsM is available as an additional pipeline within the Pcons.net protein structure prediction server. AVAILABILITY AND IMPLEMENTATION: PconsM is freely available from http://pcons.net/.     

Refinement of protein structure homology models via long, all-atom molecular dynamics simulations

Accurate computational prediction of protein structure represents a longstanding challenge in molecular biology and structure-based drug design. Although homology modeling techniques are widely used to produce low-resolution models, refining these models to high resolution has proven difficult. With long enough simulations and sufficiently accurate force fields, molecular dynamics (MD) simulations should in principle allow such refinement, but efforts to refine homology models using MD have for the most part yielded disappointing results. It has thus far been unclear whether MD-based refinement is limited primarily by accessible simulation timescales, force field accuracy, or both. Here, we examine MD as a technique for homology model refinement using all-atom simulations, each at least 100 μs long-more than 100 times longer than previous refinement simulations-and a physics-based force field that was recently shown to successfully fold a structurally diverse set of fast-folding proteins. In MD simulations of 24 proteins chosen from the refinement category of recent Critical Assessment of Structure Prediction (CASP) experiments, we find that in most cases, simulations initiated from homology models drift away from the native structure. Comparison with simulations initiated from the native structure suggests that force field accuracy is the primary factor limiting MD-based refinement. This problem can be mitigated to some extent by restricting sampling to the neighborhood of the initial model, leading to structural improvement that, while limited, is roughly comparable to the leading alternative methods.     

Template‐based protein structure modeling using TASSERVMT

Template‐based protein structure modeling is commonly used for protein structure prediction. Based on the observation that multiple template‐based methods often perform better than single template‐based methods, we further explore the use of a variable number of multiple templates for a given target in the latest variant of TASSER, TASSER^VMT. We first develop an algorithm that improves the target‐template alignment for a given template. The improved alignment, called the SP³ alternative alignment, is generated by a parametric alignment method coupled with short TASSER refinement on models selected using knowledge‐based scores. The refined top model is then structurally aligned to the template to produce the SP³ alternative alignment. Templates identified using SP³ threading are combined with the SP³ alternative and HHEARCH alignments to provide target alignments to each template. These template models are then grouped into sets containing a variable number of template/alignment combinations. For each set, we run short TASSER simulations to build full‐length models. Then, the models from all sets of templates are pooled, and the top 20–50 models selected using FTCOM ranking method. These models are then subjected to a single longer TASSER refinement run for final prediction. We benchmarked our method by comparison with our previously developed approach, pro‐sp³‐TASSER, on a set with 874 easy and 318 hard targets. The average GDT‐TS score improvements for the first model are 3.5 and 4.3% for easy and hard targets, respectively. When tested on the 112 CASP9 targets, our method improves the average GDT‐TS scores as compared to pro‐sp3‐TASSER by 8.2 and 9.3% for the 80 easy and 32 hard targets, respectively. It also shows slightly better results than the top ranked CASP9 Zhang‐Server, QUARK and HHpredA methods. The program is available for download at http://cssb.biology.gatech.edu/ . © 2011 Wiley Periodicals, Inc.     

A benchmark testing ground for integrating homology modeling and protein docking

Protein docking procedures carry out the task of predicting the structure of a protein–protein complex starting from the known structures of the individual protein components. More often than not, however, the structure of one or both components is not known, but can be derived by homology modeling on the basis of known structures of related proteins deposited in the Protein Data Bank (PDB). Thus, the problem is to develop methods that optimally integrate homology modeling and docking with the goal of predicting the structure of a complex directly from the amino acid sequences of its component proteins. One possibility is to use the best available homology modeling and docking methods. However, the models built for the individual subunits often differ to a significant degree from the bound conformation in the complex, often much more so than the differences observed between free and bound structures of the same protein, and therefore additional conformational adjustments, both at the backbone and side chain levels need to be modeled to achieve an accurate docking prediction. In particular, even homology models of overall good accuracy frequently include localized errors that unfavorably impact docking results. The predicted reliability of the different regions in the model can also serve as a useful input for the docking calculations. Here we present a benchmark dataset that should help to explore and solve combined modeling and docking problems. This dataset comprises a subset of the experimentally solved ‘target’ complexes from the widely used Docking Benchmark from the Weng Lab (excluding antibody–antigen complexes). This subset is extended to include the structures from the PDB related to those of the individual components of each complex, and hence represent potential templates for investigating and benchmarking integrated homology modeling and docking approaches. Template sets can be dynamically customized by specifying ranges in sequence similarity and in PDB release dates, or using other filtering options, such as excluding sets of specific structures from the template list. Multiple sequence alignments, as well as structural alignments of the templates to their corresponding subunits in the target are also provided. The resource is accessible online or can be downloaded at http://cluspro.org/benchmark , and is updated on a weekly basis in synchrony with new PDB releases. Proteins 2016; 85:10–16. © 2016 Wiley Periodicals, Inc.     

Development and benchmarking of TASSER(iter) for the iterative improvement of protein structure predictions

To improve the accuracy of TASSER models especially in the limit where threading provided template alignments are of poor quality, we have developed the TASSER(iter) algorithm which uses the templates and contact restraints from TASSER generated models for iterative structure refinement. We apply TASSER(iter) to a large benchmark set of 2,773 nonhomologous single domain proteins that are < or = 200 in length and that cover the PDB at the level of 35% pairwise sequence identity. Overall, TASSER(iter) models have a smaller global average RMSD of 5.48 A compared to 5.81 A RMSD of the original TASSER models. Classifying the targets by the level of prediction difficulty (where Easy targets have a good template with a corresponding good threading alignment, Medium targets have a good template but a poor alignment, and Hard targets have an incorrectly identified template), TASSER(iter) (TASSER) models have an average RMSD of 4.15 A (4.35 A) for the Easy set and 9.05 A (9.52 A) for the Hard set. The largest reduction of average RMSD is for the Medium set where the TASSER(iter) models have an average global RMSD of 5.67 A compared to 6.72 A of the TASSER models. Seventy percent of the Medium set TASSER(iter) models have a smaller RMSD than the TASSER models, while 63% of the Easy and 60% of the Hard TASSER models are improved by TASSER(iter). For the foldable cases, where the targets have a RMSD to the native <6.5 A, TASSER(iter) shows obvious improvement over TASSER models: For the Medium set, it improves the success rate from 57.0 to 67.2%, followed by the Hard targets where the success rate improves from 32.0 to 34.8%, with the smallest improvement in the Easy targets from 82.6 to 84.0%. These results suggest that TASSER(iter) can provide more reliable predictions for targets of Medium difficulty, a range that had resisted improvement in the quality of protein structure predictions.     

Refining homology models by combining replica-exchange molecular dynamics and statistical potentials

A protocol is presented for the global refinement of homology models of proteins. It combines the advantages of temperature-based replica-exchange molecular dynamics (REMD) for conformational sampling and the use of statistical potentials for model selection. The protocol was tested using 21 models. Of these 14 were models of 10 small proteins for which high-resolution crystal structures were available, the remainder were targets of the recent CASPR exercise. It was found that REMD in combination with currently available force fields could sample near-native conformational states starting from high-quality homology models. Conformations in which the backbone RMSD of secondary structure elements (SSE-RMSD) was lower than the starting value by 0.5-1.0 A were found for 15 out of the 21 cases (average 0.82 A). Furthermore, when a simple scoring function consisting of two statistical potentials was used to rank the structures, one or more structures with SSE-RMSD of at least 0.2 A lower than the starting value was found among the five best ranked structures in 11 out of the 21 cases. The average improvement in SSE-RMSD for the best models was 0.42 A. However, none of the scoring functions tested identified the structures with the lowest SSE-RMSD as the best models although all identified the native conformation as the one with lowest energy. This suggests that while the proposed protocol proved effective for the refinement of high-quality models of small proteins scoring functions remain one of the major limiting factors in structure refinement. This and other aspects by which the methodology could be further improved are discussed.     

A different look at the quality of modeled three-dimensional protein structures

Measuring the accuracy of protein three-dimensional structures is one of the most important problems in protein structure prediction. For structure-based drug design, the accuracy of the binding site is far more important than the accuracy of any other region of the protein. We have developed an automated method for assessing the quality of a protein model by focusing on the set of residues in the small molecule binding site. Small molecule binding sites typically involve multiple regions of the protein coming together in space, and their accuracy has been observed to be sensitive to even small alignment errors. In addition, ligand binding sites contain the critical information required for drug design, making their accuracy particularly important. We analyzed the accuracy of the binding sites on two sets of protein models: the predictions submitted by the top-performing CASP7 groups, and the models generated by four widely used homology modeling packages. The results of our CASP7 analysis significantly differ from the previous findings, implying that the binding site measure does not correlate with the traditional model quality measures used in the structure prediction benchmarks. For the modeling programs, the resolution of binding sites is extremely sensitive to the degree of sequence homology between the query and the template, even when the most accurate alignments are used in the homology modeling process.     

High-accuracy protein structures by combining machine-learning with physics-based refinement

Protein structure prediction has long been available as an alternative to experimental structure determination, especially via homology modeling based on templates from related sequences. Recently, models based on distance restraints from coevolutionary analysis via machine learning to have significantly expanded the ability to predict structures for sequences without templates. One such method, AlphaFold, also performs well on sequences where templates are available but without using such information directly. Here we show that combining machine-learning based models from AlphaFold with state-of-the-art physics-based refinement via molecular dynamics simulations further improves predictions to outperform any other prediction method tested during the latest round of CASP. The resulting models have highly accurate global and local structures, including high accuracy at functionally important interface residues, and they are highly suitable as initial models for crystal structure determination via molecular replacement.     

Consensus alignment for reliable framework prediction in homology modeling

MOTIVATION: Even the best sequence alignment methods frequently fail to correctly identify the framework regions for which backbones can be copied from the template into the target structure. Since the underprediction and, more significantly, the overprediction of these regions reduces the quality of the final model, it is of prime importance to attain as much as possible of the true structural alignment between target and template. RESULTS: We have developed an algorithm called Consensus that consistently provides a high quality alignment for comparative modeling. The method follows from a benchmark analysis of the 3D models generated by ten alignment techniques for a set of 79 homologous protein structure pairs. For 20-to-40% of the targets, these methods yield models with at least 6 A root mean square deviation (RMSD) from the native structure. We have selected the top five performing methods, and developed a consensus algorithm to generate an improved alignment. By building on the individual strength of each method, a set of criteria was implemented to remove the alignment segments that are likely to correspond to structurally dissimilar regions. The automated algorithm was validated on a different set of 48 protein pairs, resulting in 2.2 A average RMSD for the predicted models, and only four cases in which the RMSD exceeded 3 A. The average length of the alignments was about 75% of that found by standard structural superposition methods. The performance of Consensus was consistent from 2 to 32% target-template sequence identity, and hence it can be used for accurate prediction of framework regions in homology modeling.     

Modularity and homology: modelling of the type II module family from titin.

We report the homology modelling of the structures of the 162 type II modules from the giant multi-domain protein titin (also known as connectin). The package MODELLER was used and implemented in an automated fashion using four experimentally determined structures as templates. Validation of the models was assessed in terms of divergence from the templates and consensus of the alignments. The homology within the whole family of type II modules as well as with the templates is relatively high (20-35% identity and ca 50% similarity). Comparison between the models of domains for which an NMR structure has been solved and the experimental solution gives an estimate of the quality of the modelling. Our results allow us to distinguish between a set of structurally relevant residues, which are conserved throughout the whole family and buried in the hydrophobic core, from the residues that are conserved and exposed. These latter residues are potentially functionally important. Comparison of exposed conserved patches for modules in different regions of the titin molecule suggests potential interaction surfaces. Our results may be tested directly for those modules whose binding partner is known.     

Issues in high-throughput comparative modelling: a case study using the ubiquitin E2 conjugating enzymes

Sequences of the ubiquitin-conjugating enzyme (UBC or E2) family were used as a test set to investigate issues associated with the high-throughput comparative modelling of protein structures. A semi-automatic method was initially developed with particular emphasis on producing models of a quality suitable for structural comparison. Structural and sequence features of the E2 family were used to improve the sequence alignment and the quality of the structural templates. Initially, failure to correct for subtle structural inconsistencies between templates lead to problems in the comparative analysis of the UBC electrostatic potentials. Modelling of known UBC structures using Modeller 4.0 showed that multiple templates produced, on average, no better models than the use of just one template, as judged by the root-mean-squared deviation between the comparative model and crystal structure backbones. Using four different quality-checking methods, for a given target sequence, it was not possible to distinguish the model most similar to the experimental structure. The UBC models were thus finally modelled using only the crystal structure template with the highest sequence identity to the target to be modelled, and producing only one model solution. Quality checking was used to reject models with obvious structural anomalies (e.g., bad side-chain packing). The resulting models have been used for a comparison of UBC structural features and of their electrostatic potentials. The work was extended through the development of a fully automated pipeline that identifies E2 sequences in the sequence databases, aligns and models them, and calculates the associated electrostatic potential.     

Exploring the structure of opioid receptors with homology modeling based on single and multiple templates and subsequent docking: A comparative study

Opioid receptors are the principal targets for opioids, which have been used as analgesics for centuries. Opioid receptors belong to the rhodopsin family of G-protein coupled receptors (GPCRs). In the absence of crystal structures of opioid receptors, 3D homology models have been reported with bovine rhodopsin as a template, though the sequence homology is low. Recently, it has been reported that use of multiple templates results in a better model for a target having low sequence identity with a single template. With the objective of carrying out a comparative study on the structural quality of the 3D models based on single and multiple templates, the homology models for opioid receptors (mu, delta and kappa) were generated using bovine rhodopsin as single template and the recently deposited crystal structures of squid rhodopsin, turkey β-1 and human β-2 adrenoreceptors along with bovine rhodopsin as multiple templates. In this paper we report the results of comparison between the refined 3D models based on multiple sequence alignment (MSA) and models built with bovine rhodopsin as template, using validation programs PROCHECK, PROSA, Verify 3D, Molprobity and docking studies. The results indicate that homology models of mu and kappa with multiple templates are better than those built with only bovine rhodopsin as template, whereas, in many aspects, the homology model of delta opioid receptor with single template is better with respect to the model based on multiple templates. Three nonselective ligands were docked to both the models of mu, delta and kappa opioid receptors using GOLD 3.1. The results of docking complied well with the pharamacophore, reported for nonspecific opioid ligands. The comparison of docking results for models with multiple templates and those with single template have been discussed in detail. Three selective ligands for each receptor were also docked. As the crystallographic structures are not yet known, this comparison will help in choosing better homology models of opioid receptors for studying ligand receptor interactions to design new potent opioid antagonists.     

Improving threading algorithms for remote homology modeling by combining fragment and template comparisons

In this work, we develop a method called fragment comparison and the template comparison (FTCOM) for assessing the global quality of protein structural models for targets of medium and hard difficulty (remote homology) produced by structure prediction approaches such as threading or ab initio structure prediction. FTCOM requires the C_α coordinates of full length models and assesses model quality based on fragment comparison and a score derived from comparison of the model to top threading templates. On a set of 361 medium/hard targets, FTCOM was applied to and assessed for its ability to improve on the results from the SP³, SPARKS, PROSPECTOR_3, and PRO‐SP³‐TASSER threading algorithms. The average TM‐score improves by 5–10% for the first selected model by the new method over models obtained by the original selection procedure in the respective threading methods. Moreover, the number of foldable targets (TM‐score ≥ 0.4) increases from least 7.6% for SP³ to 54% for SPARKS. Thus, FTCOM is a promising approach to template selection. Proteins 2010. © 2010 Wiley‐Liss, Inc.     

STRUCTFAST: protein sequence remote homology detection and alignment using novel dynamic programming and profile-profile scoring

STRUCTFAST is a novel profile-profile alignment algorithm capable of detecting weak similarities between protein sequences. The increased sensitivity and accuracy of the STRUCTFAST method are achieved through several unique features. First, the algorithm utilizes a novel dynamic programming engine capable of incorporating important information from a structural family directly into the alignment process. Second, the algorithm employs a rigorous analytical formula for profile-profile scoring to overcome the limitations of ad hoc scoring functions that require adjustable parameter training. Third, the algorithm employs Convergent Island Statistics (CIS) to compute the statistical significance of alignment scores independently for each pair of sequences. STRUCTFAST routinely produces alignments that meet or exceed the quality obtained by an expert human homology modeler, as evidenced by its performance in the latest CAFASP4 and CASP6 blind prediction benchmark experiments.