Systematic and Comparative Evaluation of Software Programs for Template‐Based Modeling of Protein Structures

Modeling protein structures is critical for understanding protein functions in various biological and biotechnological studies. Among representative protein structure modeling approaches, template‐based modeling (TBM) is by far the most reliable and most widely used approach to model protein structures. However, it still remains as a challenge to select appropriate software programs for pairwise alignments and model building, two major steps of the TBM. In this paper, pairwise alignment methods for TBM are first compared with respect to the quality of structure models built using these methods. This comparative study is conducted using comprehensive datasets, which cover 6185 domain sequences from Structural Classification of Proteins extended for soluble proteins, and 259 Protein Data Bank entries (whole protein sequences) from Orientations of Proteins in Membranes database for membrane proteins. Overall, a profile‐based method, especially PSI‐BLAST, consistently shows high performance across the datasets and model evaluation metrics used. Next, use of two model building programs, MODELLER and SWISS‐MODEL, does not seem to significantly affect the quality of protein structure models built except for the Hard group (a group of relatively less homologous proteins) of membrane proteins. The results presented in this study will be useful for more accurate implementation of TBM.     

Strategies for high-throughput comparative modeling: applications to leverage analysis in structural genomics and protein family organization

The technological breakthroughs in structural genomics were designed to facilitate the solution of a sufficient number of structures, so that as many protein sequences as possible can be structurally characterized with the aid of comparative modeling. The leverage of a solved structure is the number and quality of the models that can be produced using the structure as a template for modeling and may be viewed as the "currency" with which the success of a structural genomics endeavor can be measured. Moreover, the models obtained in this way should be valuable to all biologists. To this end, at the Northeast Structural Genomics Consortium (NESG), a modular computational pipeline for automated high-throughput leverage analysis was devised and used to assess the leverage of the 186 unique NESG structures solved during the first phase of the Protein Structure Initiative (January 2000 to July 2005). Here, the results of this analysis are presented. The number of sequences in the nonredundant protein sequence database covered by quality models produced by the pipeline is approximately 39,000, so that the average leverage is approximately 210 models per structure. Interestingly, only 7900 of these models fulfill the stringent modeling criterion of being at least 30% sequence-identical to the corresponding NESG structures. This study shows how high-throughput modeling increases the efficiency of structure determination efforts by providing enhanced coverage of protein structure space. In addition, the approach is useful in refining the boundaries of structural domains within larger protein sequences, subclassifying sequence diverse protein families, and defining structure-based strategies specific to a particular family.     

Protein models docking benchmark 2

Structural characterization of protein–protein interactions is essential for our ability to understand life processes. However, only a fraction of known proteins have experimentally determined structures. Such structures provide templates for modeling of a large part of the proteome, where individual proteins can be docked by template‐free or template‐based techniques. Still, the sensitivity of the docking methods to the inherent inaccuracies of protein models, as opposed to the experimentally determined high‐resolution structures, remains largely untested, primarily due to the absence of appropriate benchmark set(s). Structures in such a set should have predefined inaccuracy levels and, at the same time, resemble actual protein models in terms of structural motifs/packing. The set should also be large enough to ensure statistical reliability of the benchmarking results. We present a major update of the previously developed benchmark set of protein models. For each interactor, six models were generated with the model‐to‐native C^α RMSD in the 1 to 6 Å range. The models in the set were generated by a new approach, which corresponds to the actual modeling of new protein structures in the “real case scenario,” as opposed to the previous set, where a significant number of structures were model‐like only. In addition, the larger number of complexes (165 vs. 63 in the previous set) increases the statistical reliability of the benchmarking. We estimated the highest accuracy of the predicted complexes (according to CAPRI criteria), which can be attained using the benchmark structures. The set is available at http://dockground.bioinformatics.ku.edu . Proteins 2015; 83:891–897. © 2015 Wiley Periodicals, Inc.     

Comparison of Composer and ORCHESTRAR

Although the number of known protein structures is increasing, the number of protein sequences without determined structures is still much larger. Three-dimensional (3D) protein structure information helps in the understanding of functional mechanisms, but solving structures by X-ray crystallography or NMR is often a lengthy and difficult process. A relatively fast way of determining a protein's 3D structure is to construct a computer model using homologous sequence and structure information. Much work has gone into algorithms that comprise the ORCHESTRAR homology modeling program in the SYBYL software package. This novel homology modeling tool combines algorithms for modeling conserved cores, variable regions, and side chains. The paradigm of using existing knowledge from multiple templates and the underlying protein environment knowledgebase is used in all of these algorithms, and will become even more powerful as the number of experimentally derived protein structures increases. To determine how ORCHESTRAR compares to Composer (a broadly used, but an older tool), homology models of 18 proteins were constructed using each program so that a detailed comparison of each step in the modeling process could be carried out. Proteins modeled include kinases, dihydrofolate reductase, HIV protease, and factor Xa. In almost all cases ORCHESTRAR produces models with lower root-mean-squared deviation (RMSD) values when compared with structures determined by X-ray crystallography or NMR. Moreover, ORCHESTRAR produced a homology model for three target sequences where Composer failed to produce any. Data for RMSD comparisons between structurally conserved cores, structurally variable regions, side-chain conformations are presented, as well as analyses of active site and protein-protein interface configurations.     

Towards fully automated structure-based function prediction in structural genomics: a case study

As the global Structural Genomics projects have picked up pace, the number of structures annotated in the Protein Data Bank as hypothetical protein or unknown function has grown significantly. A major challenge now involves the development of computational methods to assign functions to these proteins accurately and automatically. As part of the Midwest Center for Structural Genomics (MCSG) we have developed a fully automated functional analysis server, ProFunc, which performs a battery of analyses on a submitted structure. The analyses combine a number of sequence-based and structure-based methods to identify functional clues. After the first stage of the Protein Structure Initiative (PSI), we review the success of the pipeline and the importance of structure-based function prediction. As a dataset, we have chosen all structures solved by the MCSG during the 5 years of the first PSI. Our analysis suggests that two of the structure-based methods are particularly successful and provide examples of local similarity that is difficult to identify using current sequence-based methods. No one method is successful in all cases, so, through the use of a number of complementary sequence and structural approaches, the ProFunc server increases the chances that at least one method will find a significant hit that can help elucidate function. Manual assessment of the results is a time-consuming process and subject to individual interpretation and human error. We present a method based on the Gene Ontology (GO) schema using GO-slims that can allow the automated assessment of hits with a success rate approaching that of expert manual assessment.     

Refinement of homology-based protein structures by molecular dynamics simulation techniques

The use of classical molecular dynamics simulations, performed in explicit water, for the refinement of structural models of proteins generated ab initio or based on homology has been investigated. The study involved a test set of 15 proteins that were previously used by Baker and coworkers to assess the efficiency of the ROSETTA method for ab initio protein structure prediction. For each protein, four models generated using the ROSETTA procedure were simulated for periods of between 5 and 400 nsec in explicit solvent, under identical conditions. In addition, the experimentally determined structure and the experimentally derived structure in which the side chains of all residues had been deleted and then regenerated using the WHATIF program were simulated and used as controls. A significant improvement in the deviation of the model structures from the experimentally determined structures was observed in several cases. In addition, it was found that in certain cases in which the experimental structure deviated rapidly from the initial structure in the simulations, indicating internal strain, the structures were more stable after regenerating the side-chain positions. Overall, the results indicate that molecular dynamics simulations on a tens to hundreds of nanoseconds time scale are useful for the refinement of homology or ab initio models of small to medium-size proteins.     

1,000 structures and more from the MCSG

Background  

The Midwest Center for Structural Genomics (MCSG) is one of the large-scale centres of the Protein Structure Initiative (PSI). During the first two phases of the PSI the MCSG has solved over a thousand protein structures. A criticism of structural genomics is that target selection strategies mean that some structures are solved without having a known function and thus are of little biomedical significance. Structures of unknown function have stimulated the development of methods for function prediction from structure.     

Protein models: The Grand Challenge of protein docking

Characterization of life processes at the molecular level requires structural details of protein–protein interactions (PPIs). The number of experimentally determined protein structures accounts only for a fraction of known proteins. This gap has to be bridged by modeling, typically using experimentally determined structures as templates to model related proteins. The fraction of experimentally determined PPI structures is even smaller than that for the individual proteins, due to a larger number of interactions than the number of individual proteins, and a greater difficulty of crystallizing protein–protein complexes. The approaches to structural modeling of PPI (docking) often have to rely on modeled structures of the interactors, especially in the case of large PPI networks. Structures of modeled proteins are typically less accurate than the ones determined by X‐ray crystallography or nuclear magnetic resonance. Thus the utility of approaches to dock these structures should be assessed by thorough benchmarking, specifically designed for protein models. To be credible, such benchmarking has to be based on carefully curated sets of structures with levels of distortion typical for modeled proteins. This article presents such a suite of models built for the benchmark set of the X‐ray structures from the Dockground resource ( http://dockground.bioinformatics.ku.edu ) by a combination of homology modeling and Nudged Elastic Band method. For each monomer, six models were generated with predefined C^α root mean square deviation from the native structure (1, 2, …, 6 Å). The sets and the accompanying data provide a comprehensive resource for the development of docking methodology for modeled proteins. Proteins 2014; 82:278–287. © 2013 Wiley Periodicals, Inc.     

High-quality homology models derived from NMR and X-ray structures of E. coli proteins YgdK and Suf E suggest that all members of the YgdK/Suf E protein family are enhancers of cysteine desulfurases

The structural biology of proteins mediating iron-sulfur (Fe-S) cluster assembly is central for understanding several important biological processes. Here we present the NMR structure of the 16-kDa protein YgdK from Escherichia coli, which shares 35% sequence identity with the E. coli protein SufE. The SufE X-ray crystal structure was solved in parallel with the YdgK NMR structure in the Northeast Structural Genomics (NESG) consortium. Both proteins are (1) key components for Fe-S metabolism, (2) exhibit the same distinct fold, and (3) belong to a family of at least 70 prokaryotic and eukaryotic sequence homologs. Accurate homology models were calculated for the YgdK/SufE family based on YgdK NMR and SufE crystal structure. Both structural templates contributed equally, exemplifying synergy of NMR and X-ray crystallography. SufE acts as an enhancer of the cysteine desulfurase activity of SufS by SufE-SufS complex formation. A homology model of CsdA, a desulfurase encoded in the same operon as YgdK, was modeled using the X-ray structure of SufS as a template. Protein surface and electrostatic complementarities strongly suggest that YgdK and CsdA likewise form a functional two-component desulfurase complex. Moreover, structural features of YgdK and SufS, which can be linked to their interaction with desulfurases, are conserved in all homology models. It thus appears very likely that all members of the YgdK/SufE family act as enhancers of Suf-S-like desulfurases. The present study exemplifies that "refined" selection of two (or more) targets enables high-quality homology modeling of large protein families.     

Functional differentiation of proteins: implications for structural genomics

Structural genomics is a broad initiative of various centers aiming to provide complete coverage of protein structure space. Because it is not feasible to experimentally determine the structures of all proteins, it is generally agreed that the only viable strategy to achieve such coverage is to carefully select specific proteins (targets), determine their structure experimentally, and then use comparative modeling techniques to model the rest. Here we suggest that structural genomics centers refine the structure-driven approach in target selection by adopting function-based criteria. We suggest targeting functionally divergent superfamilies within a given structural fold so that each function receives a structural characterization. We have developed a method to do so, and an itemized survey of several functionally rich folds shows that they are only partially functionally characterized. We call upon structural genomics centers to consider this approach and upon computational biologists to further develop function-based targeting methods.     

High-throughput computational structure-based characterization of protein families: START domains and implications for structural genomics

SkyLine, a high-throughput homology modeling pipeline tool, detects and models true sequence homologs to a given protein structure. Structures and models are stored in SkyBase with links to computational function annotation, as calculated by MarkUs. The SkyLine/SkyBase/MarkUs technology represents a novel structure-based approach that is more objective and versatile than other protein classification resources. This structure-centric strategy provides a multi-dimensional organization and coverage of protein space at the levels of family, function, and genome. The concept of “modelability”, the ability to model sequences on related structures, provides a reliable criterion for membership in a protein family (“leverage”) and underlies the unique success of this approach. The overall procedure is illustrated by its application to START domains, which comprise a Biomedical Theme for the Northeast Structural Genomics Consortium as part of the Protein Structure Initiative. START domains are typically involved in the non-vesicular transport of lipids. While 19 experimentally determined structures are available, the family, whose evolutionary hierarchy is not well determined, is highly sequence diverse, and the ligand-binding potential of many family members is unknown. The SkyLine/SkyBase/MarkUs approach provides significant insights and predicts: (1) many more family members (~4,000) than any other resource; (2) the function for a large number of unannotated proteins; (3) instances of START domains in genomes from which they were thought to be absent; and (4) the existence of two types of novel proteins, those containing dual START domain and those containing N-terminal START domains.     

Assessment of the ability to model proteins with leucine-rich repeats in light of the latest structural information

The three-dimensional structures of leucine-rich repeat (LRR)-containing proteins from five different families were previously predicted based on the crystal structure of the ribonuclease inhibitor, using an approach that combined homology-based modeling, structure-based sequence alignment of LRRs, and several rational assumptions. The structural models have been produced based on very limited sequence similarity, which, in general, cannot yield trustworthy predictions. Recently, the protein structures from three of these five families have been determined. In this report we estimate the quality of the modeling approach by comparing the models with the experimentally determined structures. The comparison suggests that the general architecture, curvature, "interior/exterior" orientations of side chains, and backbone conformation of the LRR structures can be predicted correctly. On the other hand, the analysis revealed that, in some cases, it is difficult to predict correctly the twist of the overall super-helical structure. Taking into consideration the conclusions from these comparisons, we identified a new family of bacterial LRR proteins and present its structural model. The reliability of the LRR protein modeling suggests that it would be informative to apply similar modeling approaches to other classes of solenoid proteins.     

Molecular modeling of protein function regions

Experimental protein structures often provide extensive insight into the mode and specificity of small molecule binding, and this information is useful for understanding protein function and for the design of drugs. We have performed an analysis of the reliability with which ligand-binding information can be deduced from computer model structures, as opposed to experimentally derived ones. Models produced as part of the CASP experiments are used. The accuracy of contacts between protein model atoms and experimentally determined ligand atom positions is the main criterion. Only comparative models are included (i.e., models based on a sequence relationship between the protein of interest and a known structure). We find that, as expected, contact errors increase with decreasing sequence identity used as a basis for modeling. Analysis of the causes of errors shows that sequence alignment errors between model and experimental template have the most deleterious effect. In general, good, but not perfect, insight into ligand binding can be obtained from models based on a sequence relationship, providing there are no alignment errors in the model. The results support a structural genomics strategy based on experimental sampling of structure space so that all protein domains can be modeled on the basis of 30% or higher sequence identity.     

The structure of the infectious prion protein

The structures of the infectious prion protein, PrP^Sc, and that of its proteolytically truncated variant, PrP 27–30, have evaded experimental determination due to their insolubility and propensity to aggregate. Molecular modeling has been used to fill this void and to predict their structures, but various modeling approaches have produced significantly different models. The disagreement between the different modeling solutions indicates the limitations of this method. Over the years, in absence of a three-dimensional (3D) structure, a variety of experimental techniques have been used to gain insights into the structure of this biologically, medically, and agriculturally important isoform. Here, we present an overview of experimental results that were published in recent years, and which provided new insights into the molecular architecture of PrP^Sc and PrP 27–30. Furthermore, we evaluate all published models in light of these recent, experimental data, and come to the conclusion that none of the models can accommodate all of the experimental constraints. Moreover, this conclusion constitutes an open invitation for renewed efforts to model the structure of PrP^Sc.     

Assessment of 3D models for allergen research

Allergenic proteins must crosslink specific IgE molecules, bound to the surface of mast cells and basophils, to stimulate an immune response. A structural understanding of the allergen–IgE interface is needed to predict cross‐reactivities between allergens and to design hypoallergenic proteins. However, there are less than 90 experimentally determined structures available for the approximately 1500 sequences of allergens and isoallergens cataloged in the Structural Database of Allergenic Proteins. To provide reliable structural data for the remaining proteins, we previously produced more than 500 3D models using an automated procedure, with strict controls on template choice and model quality evaluation. Here, we assessed how well the fold and residue surface exposure of 10 of these models correlated with recently published experimental 3D structures determined by X‐ray crystallography or NMR. We also discuss the impact of intrinsically disordered regions on the structural comparison and epitope prediction. Overall, for seven allergens with sequence identities to the original templates higher than 27%, the backbone root‐mean square deviations were less than 2 Å between the models and the subsequently determined experimental structures for the ordered regions. Further, the surface exposure of the known IgE epitopes on the models of three major allergens, from peanut (Ara h 1), latex (Hev b 2), and soy (Gly m 4), was very similar to the experimentally determined structures. For the three remaining allergens with lower sequence identities to the modeling templates, the 3D folds were correctly identified. However, the accuracy of those models is not sufficient for a reliable epitope mapping. © Proteins 2013. © 2012 Wiley Periodicals, Inc.     

Structure-based functional inference in structural genomics

The dramatically increasing number of new protein sequences arising from genomics 4 proteomics requires the need for methods to rapidly and reliably infer the molecular and cellular functions of these proteins. One such approach, structural genomics, aims to delineate the total repertoire of protein folds in nature, thereby providing three-dimensional folding patterns for all proteins and to infer molecular functions of the proteins based on the combined information of structures and sequences. The goal of obtaining protein structures on a genomic scale has motivated the development of high throughput technologies and protocols for macromolecular structure determination that have begun to produce structures at a greater rate than previously possible. These new structures have revealed many unexpected functional inferences and evolutionary relationships that were hidden at the sequence level. Here, we present samples of structures determined at Berkeley Structural Genomics Center and collaborators laboratories to illustrate how structural information provides and complements sequence information to deduce the functional inferences of proteins with unknown molecular functions.Two of the major premises of structural genomics are to discover a complete repertoire of protein folds in nature and to find molecular functions of the proteins whose functions are not predicted from sequence comparison alone. To achieve these objectives on a genomic scale, new methods, protocols, and technologies need to be developed by multi-institutional collaborations worldwide. As part of this effort, the Protein Structure Initiative has been launched in the United States (PSI; www.nigms.nih.gov/funding/psi.html). Although infrastructure building and technology development are still the main focus of structural genomics programs [1–6], a considerable number of protein structures have already been produced, some of them coming directly out of semi-automated structure determination pipelines [6–10]. The Berkeley Structural Genomics Center (BSGC) has focused on the proteins of Mycoplasma or their homologues from other organisms as its structural genomics targets because of the minimal genome size of the Mycoplasmas as well as their relevance to human and animal pathogenicity (http://www.strgen.org). Here we present several protein examples encompassing a spectrum of functional inferences obtainable from their three-dimensional structures in five situations, where the inferences are new and testable, and are not predictable from protein sequence information alone.     

MODBASE, a database of annotated comparative protein structure models

MODBASE (http://guitar.rockefeller.edu/modbase) is a relational database of annotated comparative protein structure models for all available protein sequences matched to at least one known protein structure. The models are calculated by MODPIPE, an automated modeling pipeline that relies on PSI-BLAST, IMPALA and MODELLER. MODBASE uses the MySQL relational database management system for flexible and efficient querying, and the MODVIEW Netscape plugin for viewing and manipulating multiple sequences and structures. It is updated regularly to reflect the growth of the protein sequence and structure databases, as well as improvements in the software for calculating the models. For ease of access, MODBASE is organized into different datasets. The largest dataset contains models for domains in 304 517 out of 539 171 unique protein sequences in the complete TrEMBL database (23 March 2001); only models based on significant alignments (PSI-BLAST E-value < 10^–4) and models assessed to have the correct fold are included. Other datasets include models for target selection and structure-based annotation by the New York Structural Genomics Research Consortium, models for prediction of genes in the Drosophila melanogaster genome, models for structure determination of several ribosomal particles and models calculated by the MODWEB comparative modeling web server.     

Molecular interactions between photosystem I and ferredoxin: an integrated energy frustration and experimental model

The stromal domain (PsaC, PsaD, and PsaE) of photosystem I (PSI) reduces transiently bound ferredoxin (Fd) or flavodoxin. Experimental structures exist for all of these protein partners individually, but no experimental structure of the PSI/Fd or PSI/flavodoxin complexes is presently available. Molecular models of Fd docked onto the stromal domain of the cyanobacterial PSI site are constructed here utilizing X‐ray and NMR structures of PSI and Fd, respectively. Predictions of potential protein‐protein interaction regions are based on experimental site‐directed mutagenesis and cross‐linking studies to guide rigid body docking calculations of Fd into PSI, complemented by energy landscape theory to bring together regions of high energetic frustration on each of the interacting proteins. The results identify two regions of high localized frustration on the surface of Fd that contain negatively charged Asp and Glu residues. This study predicts that these regions interact predominantly with regions of high localized frustration on the PsaC, PsaD, and PsaE chains of PSI, which include several residues predicted by previous experimental studies. Copyright © 2014 John Wiley & Sons, Ltd.