A comparative study of available software for high-accuracy homology modeling: from sequence alignments to structural models

An open question in protein homology modeling is, how well do current modeling packages satisfy the dual criteria of quality of results and practical ease of use? To address this question objectively, we examined homology‐built models of a variety of therapeutically relevant proteins. The sequence identities across these proteins range from 19% to 76%. A novel metric, the difference alignment index (DAI), is developed to aid in quantifying the quality of local sequence alignments. The DAI is also used to construct the relative sequence alignment (RSA), a new representation of global sequence alignment that facilitates comparison of sequence alignments from different methods. Comparisons of the sequence alignments in terms of the RSA and alignment methodologies are made to better understand the advantages and caveats of each method. All sequence alignments and corresponding 3D models are compared to their respective structure‐based alignments and crystal structures. A variety of protein modeling software was used. We find that at sequence identities >40%, all packages give similar (and satisfactory) results; at lower sequence identities (<25%), the sequence alignments generated by Profit and Prime, which incorporate structural information in their sequence alignment, stand out from the rest. Moreover, the model generated by Prime in this low sequence identity region is noted to be superior to the rest. Additionally, we note that DSModeler and MOE, which generate reasonable models for sequence identities >25%, are significantly more functional and easier to use when compared with the other structure‐building software.     

A rapid protein structure alignment algorithm based on a text modeling technique

Structural alignment of proteins is widely used in various fields of structural biology. In order to further improve the quality of alignment, we describe an algorithm for structural alignment based on text modelling techniques. The technique firstly superimposes secondary structure elements of two proteins and then, models the 3D-structure of the protein in a sequence of alphabets. These sequences are utilized by a step-by-step sequence alignment procedure to align two protein structures. A benchmark test was organized on a set of 200 non-homologous proteins to evaluate the program and compare it to state of the art programs, e.g. CE, SAL, TM-align and 3D-BLAST. On average, the results of all-against-all structure comparison by the program have a competitive accuracy with CE and TM-align where the algorithm has a high running speed like 3D-BLAST.     

Local alignment refinement using structural assessment

Homology modeling is the most commonly used technique to build a three-dimensional model for a protein sequence. It heavily relies on the quality of the sequence alignment between the protein to model and related proteins with a known three dimensional structure. Alignment quality can be assessed according to the physico-chemical properties of the three dimensional models it produces. In this work, we introduce fifteen predictors designed to evaluate the properties of the models obtained for various alignments. They consist of an energy value obtained from different force fields (CHARMM, ProsaII or ANOLEA) computed on residue selected around misaligned regions. These predictors were evaluated on ten challenging test cases. For each target, all possible ungapped alignments are generated and their corresponding models are computed and evaluated. The best predictor, retrieving the structural alignment for 9 out of 10 test cases, is based on the ANOLEA atomistic mean force potential and takes into account residues around misaligned secondary structure elements. The performance of the other predictors is significantly lower. This work shows that substantial improvement in local alignments can be obtained by careful assessment of the local structure of the resulting models.     

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.     

Improving the quality of twilight-zone alignments

Several recent publications illustrated advantages of using sequence profiles in recognizing distant homologies between proteins. At the same time, the practical usefulness of distant homology recognition depends not only on the sensitivity of the algorithm, but also on the quality of the alignment between a prediction target and the template from the database of known proteins. Here, we study this question for several supersensitive protein algorithms that were previously compared in their recognition sensitivity (Rychlewski et al., 2000). A database of protein pairs with similar structures, but low sequence similarity is used to rate the alignments obtained with several different methods, which included sequence-sequence, sequence-profile, and profile-profile alignment methods. We show that incorporation of evolutionary information encoded in sequence profiles into alignment calculation methods significantly increases the alignment accuracy, bringing them closer to the alignments obtained from structure comparison. In general, alignment quality is correlated with recognition and alignment score significance. For every alignment method, alignments with statistically significant scores correlate with both correct structural templates and good quality alignments. At the same time, average alignment lengths differ in various methods, making the comparison between them difficult. For instance, the alignments obtained by FFAS, the profile-profile alignment algorithm developed in our group are always longer that the alignments obtained with the PSI-BLAST algorithms. To address this problem, we develop methods to truncate or extend alignments to cover a specified percentage of protein lengths. In most cases, the elongation of the alignment by profile-profile methods is reasonable, adding fragments of similar structure. The examples of erroneous alignment are examined and it is shown that they can be identified based on the model quality.     

Multiple protein sequence alignment from tertiary structure comparison: assignment of global and residue confidence levels.

An algorithm is presented for the accurate and rapid generation of multiple protein sequence alignments from tertiary structure comparisons. A preliminary multiple sequence alignment is performed using sequence information, which then determines an initial superposition of the structures. A structure comparison algorithm is applied to all pairs of proteins in the superimposed set and a similarity tree calculated. Multiple sequence alignments are then generated by following the tree from the branches to the root. At each branchpoint of the tree, a structure-based sequence alignment and coordinate transformations are output, with the multiple alignment of all structures output at the root. The algorithm encoded in STAMP (STructural Alignment of Multiple Proteins) is shown to give alignments in good agreement with published structural accounts within the dehydrogenase fold domains, globins, and serine proteinases. In order to reduce the need for visual verification, two similarity indices are introduced to determine the quality of each generated structural alignment. Sc quantifies the global structural similarity between pairs or groups of proteins, whereas Pij' provides a normalized measure of the confidence in the alignment of each residue. STAMP alignments have the quality of each alignment characterized by Sc and Pij' values and thus provide a reproducible resource for studies of residue conservation within structural motifs.     

MUMMALS: multiple sequence alignment improved by using hidden Markov models with local structural information

We have developed MUMMALS, a program to construct multiple protein sequence alignment using probabilistic consistency. MUMMALS improves alignment quality by using pairwise alignment hidden Markov models (HMMs) with multiple match states that describe local structural information without exploiting explicit structure predictions. Parameters for such models have been estimated from a large library of structure-based alignments. We show that (i) on remote homologs, MUMMALS achieves statistically best accuracy among several leading aligners, such as ProbCons, MAFFT and MUSCLE, albeit the average improvement is small, in the order of several percent; (ii) a large collection (>10000) of automatically computed pairwise structure alignments of divergent protein domains is superior to smaller but carefully curated datasets for estimation of alignment parameters and performance tests; (iii) reference-independent evaluation of alignment quality using sequence alignment-dependent structure superpositions correlates well with reference-dependent evaluation that compares sequence-based alignments to structure-based reference alignments.     

Protein sequence threading: Averaging over structures

Multiple sequence alignments are a routine tool in protein fold recognition, but multiple structure alignments are computationally less cooperative. This work describes a method for protein sequence threading and sequence-to-structure alignments that uses multiple aligned structures, the aim being to improve models from protein threading calculations. Sequences are aligned into a field due to corresponding sites in homologous proteins. On the basis of a test set of more than 570 protein pairs, the procedure does improve alignment quality, although no more than averaging over sequences. For the force field tested, the benefit of structure averaging is smaller than that of adding sequence similarity terms or a contribution from secondary structure predictions. Although there is a significant improvement in the quality of sequence-to-structure alignments, this does not directly translate to an immediate improvement in fold recognition capability.     

A segment alignment approach to protein comparison

MOTIVATION: Local structure segments (LSSs) are small structural units shared by unrelated proteins. They are extensively used in protein structure comparison, and predicted LSSs (PLSSs) are used very successfully in ab initio folding simulations. However, predicted or real LSSs are rarely exploited by protein sequence comparison programs that are based on position-by-position alignments. RESULTS: We developed a SEgment Alignment algorithm (SEA) to compare proteins described as a collection of predicted local structure segments (PLSSs), which is equivalent to an unweighted graph (network). Any specific structure, real or predicted corresponds to a specific path in this network. SEA then uses a network matching approach to find two most similar paths in networks representing two proteins. SEA explores the uncertainty and diversity of predicted local structure information to search for a globally optimal solution. It simultaneously solves two related problems: the alignment of two proteins and the local structure prediction for each of them. On a benchmark of protein pairs with low sequence similarity, we show that application of the SEA algorithm improves alignment quality as compared to FFAS profile-profile alignment, and in some cases SEA alignments can match the structural alignments, a feat previously impossible for any sequence based alignment methods.     

Relationship between multiple sequence alignments and quality of protein comparative models

Comparative modeling is the method of choice, whenever applicable, for protein structure prediction, not only because of its higher accuracy compared to alternative methods, but also because it is possible to estimate a priori the quality of the models that it can produce, thereby allowing the usefulness of a model for a given application to be assessed beforehand. By and large, the quality of a comparative model depends on two factors: the extent of structural divergence between the target and the template and the quality of the sequence alignment between the two protein sequences. The latter is usually derived from a multiple sequence alignment (MSA) of as many proteins of the family as possible, and its accuracy depends on the number and similarity distribution of the sequences of the protein family. Here we describe a method to evaluate the expected difficulty, and by extension accuracy, of a comparative model on the basis of the MSA used to build it. The parameter that we derive is used to compare the results obtained in the last two editions of the Critical Assessment of Methods for Structure Prediction (CASP) experiment as a function of the difficulty of the modeling exercise. Our analysis demonstrates that the improvement in the scope and quality of comparative models between the two experiments is largely due to the increased number of available protein sequences and to the consequent increased chance that a large and appropriately spaced set of protein sequences homologous to the proteins of interest is available.     

An integrated approach to the analysis and modeling of protein sequences and structures. III. A comparative study of sequence conservation in protein structural families using multiple structural alignments

The information required to generate a protein structure is contained in its amino acid sequence, but how three-dimensional information is mapped onto a linear sequence is still incompletely understood. Multiple structure alignments of similar protein structures have been used to investigate conserved sequence features but contradictory results have been obtained, due, in large part, to the absence of subjective criteria to be used in the construction of sequence profiles and in the quantitative comparison of alignment results. Here, we report a new procedure for multiple structure alignment and use it to construct structure-based sequence profiles for similar proteins. The definition of "similar" is based on the structural alignment procedure and on the protein structural distance (PSD) described in paper I of this series, which offers an objective measure for protein structure relationships. Our approach is tested in two well-studied groups of proteins; serine proteases and Ig-like proteins. It is demonstrated that the quality of a sequence profile generated by a multiple structure alignment is quite sensitive to the PSD used as a threshold for the inclusion of proteins in the alignment. Specifically, if the proteins included in the aligned set are too distant in structure from one another, there will be a dilution of information and patterns that are relevant to a subset of the proteins are likely to be lost.In order to understand better how the same three-dimensional information can be encoded in seemingly unrelated sequences, structure-based sequence profiles are constructed for subsets of proteins belonging to nine superfolds. We identify patterns of relatively conserved residues in each subset of proteins. It is demonstrated that the most conserved residues are generally located in the regions where tertiary interactions occur and that are relatively conserved in structure. Nevertheless, the conservation patterns are relatively weak in all cases studied, indicating that structure-determining factors that do not require a particular sequential arrangement of amino acids, such as secondary structure propensities and hydrophobic interactions, are important in encoding protein fold information. In general, we find that similar structures can fold without having a set of highly conserved residue clusters or a well-conserved sequence profile; indeed, in some cases there is no apparent conservation pattern common to structures with the same fold. Thus, when a group of proteins exhibits a common and well-defined sequence pattern, it is more likely that these sequences have a close evolutionary relationship rather than the similarities having arisen from the structural requirements of a given fold.     

More for less in structural genomics

Structural genomics is the idea of covering protein space so that every protein sequence comes within model building distance of a protein of known structure. Unfortunately, reproducing the structural alignment of distantly related proteins is a difficult challenge to existing sequence alignment and motif search software. We have developed a new transitive alignment algorithm (MaxFlow), which generates accurate alignments between proteins deep in the twilight zone of sequence similarity, below 20% sequence identity. In particular, MaxFlow reliably identifies conserved core motifs between proteins which are only indirect PSI-Blast neighbours. Based on MaxFlow alignments, useful 3D models can be generated for all members of a superfamily from as few as a single structural template – despite hundreds of representatives at 40% sequence identity level and patchy detection of homology by PSI-Blast. We propose novel strategies for target prioritization using MaxFlow scores to predict the optimal templates in a superfamily. Our results support an increase in the granularity of covering protein space that has potentially enormous economic implications for planning the transition to the full production phase of structural genomics.     

A novel approach to remote homology detection: jumping alignments.

We describe a new algorithm for protein classification and the detection of remote homologs. The rationale is to exploit both vertical and horizontal information of a multiple alignment in a well-balanced manner. This is in contrast to established methods such as profiles and profile hidden Markov models which focus on vertical information as they model the columns of the alignment independently and to family pairwise search which focuses on horizontal information as it treats given sequences separately. In our setting, we want to select from a given database of "candidate sequences" those proteins that belong to a given superfamily. In order to do so, each candidate sequence is separately tested against a multiple alignment of the known members of the superfamily by means of a new jumping alignment algorithm. This algorithm is an extension of the Smith-Waterman algorithm and computes a local alignment of a single sequence and a multiple alignment. In contrast to traditional methods, however, this alignment is not based on a summary of the individual columns of the multiple alignment. Rather, the candidate sequence is at each position aligned to one sequence of the multiple alignment, called the "reference sequence." In addition, the reference sequence may change within the alignment, while each such jump is penalized. To evaluate the discriminative quality of the jumping alignment algorithm, we compare it to profiles, profile hidden Markov models, and family pairwise search on a subset of the SCOP database of protein domains. The discriminative quality is assessed by median false positive counts (med-FP-counts). For moderate med-FP-counts, the number of successful searches with our method is considerably higher than with the competing methods.     

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.     

Globin-like蛋白质折叠类型识别

蛋白质折叠类型识别是蛋白质结构研究的重要内容.以SCOP中的Globin-like折叠为研究对象,选择其中序列同一性小于25%的17个代表性蛋白质为训练集,采用机器和人工结合的办法进行结构比对,产生序列排比,经过训练得到了适合Globin-like折叠的概形隐马尔科夫模型(profile HMM)用于该折叠类型的识别.以Astrall.65中的68057个结构域样本进行检验,识别敏感度为99.64%,特异性100%.在折叠类型水平上,与Pfam和SUPERFAMILY单纯使用序列比对构建的HMM相比,所用模型由多于100个归为一个,仍然保持了很高的识别效果.结果表明:对序列相似度很低但具有相同折叠类型的蛋白质,可以通过引入结构比对的方法建立统一的HMM模型,实现高准确率的折叠类型识别.     

Multiple Model Approach: Exploring the Limits of Comparative Modeling

One of the biggest problems in modeling distantly related proteins is the quality of the target-template alignment. This problem often results in low quality models that do not utilize all the information available in the template structure. The divergence of alignments at a low sequence identity level, which is a hindrance in most modeling attempts, is used here as a basis for a new technique of Multiple Model Approach (MMA). Alternative alignments prepared here using different mutation matrices and gap penalties, combined with automated model building, are used to create a set of models that explore a range of possible conformations for the target protein. Models are evaluated using different techniques to identify the best model. In the set of examples studied here, the correct target structure is known, which allows the evaluation of various alignment and evaluation strategies. For a randomly selected group of distantly homologous protein pairs representing all structural classes and various fold types, it is shown that a threading score based on simplified statistical potentials of mean force can identify the best models and, consequently, the most reliable alignment. In cases where the difference between target and template structures is significant, the threading score shows clearly that all models are wrong, therefore disqualifying the template.     

GeneSilico protein structure prediction meta-server

Rigorous assessments of protein structure prediction have demonstrated that fold recognition methods can identify remote similarities between proteins when standard sequence search methods fail. It has been shown that the accuracy of predictions is improved when refined multiple sequence alignments are used instead of single sequences and if different methods are combined to generate a consensus model. There are several meta-servers available that integrate protein structure predictions performed by various methods, but they do not allow for submission of user-defined multiple sequence alignments and they seldom offer confidentiality of the results. We developed a novel WWW gateway for protein structure prediction, which combines the useful features of other meta-servers available, but with much greater flexibility of the input. The user may submit an amino acid sequence or a multiple sequence alignment to a set of methods for primary, secondary and tertiary structure prediction. Fold-recognition results (target-template alignments) are converted into full-atom 3D models and the quality of these models is uniformly assessed. A consensus between different FR methods is also inferred. The results are conveniently presented on-line on a single web page over a secure, password-protected connection. The GeneSilico protein structure prediction meta-server is freely available for academic users at http://genesilico.pl/meta.     

On the accuracy of homology modeling and sequence alignment methods applied to membrane proteins

In this study, we investigate the extent to which techniques for homology modeling that were developed for water-soluble proteins are appropriate for membrane proteins as well. To this end we present an assessment of current strategies for homology modeling of membrane proteins and introduce a benchmark data set of homologous membrane protein structures, called HOMEP. First, we use HOMEP to reveal the relationship between sequence identity and structural similarity in membrane proteins. This analysis indicates that homology modeling is at least as applicable to membrane proteins as it is to water-soluble proteins and that acceptable models (with C alpha-RMSD values to the native of 2 A or less in the transmembrane regions) may be obtained for template sequence identities of 30% or higher if an accurate alignment of the sequences is used. Second, we show that secondary-structure prediction algorithms that were developed for water-soluble proteins perform approximately as well for membrane proteins. Third, we provide a comparison of a set of commonly used sequence alignment algorithms as applied to membrane proteins. We find that high-accuracy alignments of membrane protein sequences can be obtained using state-of-the-art profile-to-profile methods that were developed for water-soluble proteins. Improvements are observed when weights derived from the secondary structure of the query and the template are used in the scoring of the alignment, a result which relies on the accuracy of the secondary-structure prediction of the query sequence. The most accurate alignments were obtained using template profiles constructed with the aid of structural alignments. In contrast, a simple sequence-to-sequence alignment algorithm, using a membrane protein-specific substitution matrix, shows no improvement in alignment accuracy. We suggest that profile-to-profile alignment methods should be adopted to maximize the accuracy of homology models of membrane proteins.