ClusPro in rounds 38 to 45 of CAPRI: Toward combining template-based methods with free docking

Targets in the protein docking experiment CAPRI (Critical Assessment of Predicted Interactions) generally present new challenges and contribute to new developments in methodology. In rounds 38 to 45 of CAPRI, most targets could be effectively predicted using template-based methods. However, the server ClusPro required structures rather than sequences as input, and hence we had to generate and dock homology models. The available templates also provided distance restraints that were directly used as input to the server. We show here that such an approach has some advantages. Free docking with template-based restraints using ClusPro reproduced some interfaces suggested by weak or ambiguous templates while not reproducing others, resulting in correct server predicted models. More recently we developed the fully automated ClusPro TBM server that performs template-based modeling and thus can use sequences rather than structures of component proteins as input. The performance of the server, freely available for noncommercial use at https://tbm.cluspro.org , is demonstrated by predicting the protein-protein targets of rounds 38 to 45 of CAPRI.     

Predicting oligomeric assemblies: N-mers a primer

Multi-protein complexes play key roles in many biological processes. However, since the structures of these assemblies are hard to resolve experimentally, the detailed mechanism of how they work cooperatively in the cell has remained elusive. Similarly, recent advances on in silico prediction of protein-protein interactions have so far avoided this difficult problem. In this paper, we present a general algorithm to predict molecular assemblies of homo-oligomers. Given the number of N-mers and the 3D structure of one monomer, the method samples all the possible symmetries that N-mers can be assembled. Based on a scoring function that clusters the low free energy structures at each binding interface, the algorithm predicts the complex structure as well as the symmetry of the protein assembly. The method is quite general and does not involve any free parameters. The algorithm has been implemented as a public server and integrated to the protein-protein complex prediction server ClusPro. Using this application, we validated predictions for trimers, tetramers (discriminating between dimer of dimers and 4-fold symmetry structures), pentamers and hexamers (discriminating between trimer of dimers, dimer of trimers, and 6-fold symmetry structures), for a total of 107 assemblies. For 85% of the multimers, the server predicts the complex structure within an average rms deviation of 2A from the full crystal. For complexes that involve more than one binding interface, the cluster size at each surface provides a strong indication as to which interface forms first. With improving scoring functions and computer power, our multimer docking approach could be used as a framework to address the more general problem of multi-protein assemblies.     

Modeling side-chains using molecular dynamics improve recognition of binding region in CAPRI targets

The CAPRI-II experiment added an extra level of complexity to the problem of predicting protein-protein interactions by including 5 targets for which participants had to build or complete the 3-dimensional (3D) structure of either the receptor or ligand based on the structure of a close homolog. In this article, we describe how modeling key side-chains using molecular dynamics (MD) in explicit solvent improved the recognition of the binding region of a free energy- based computational docking method. In particular, we show that MD is able to predict with relatively high accuracy the rotamer conformation of the anchor side-chains important for molecular recognition as suggested by Rajamani et al. (Proc Natl Acad Sci USA 2004;101:11287-11292). As expected, the conformations are some of the most common rotamers for the given residue, while latch side-chains that undergo induced fit upon binding are forced into less common conformations. Using these models as starting conformations in conjunction with the rigid-body docking server ClusPro and the flexible docking algorithm SmoothDock, we produced valuable predictions for 6 of the 9 targets in CAPRI-II, missing only the 3 targets that underwent significant structural rearrangements upon binding. We also show that our free energy- based scoring function, consisting of the sum of van der Waals, Coulombic electrostatic with a distance-dependent dielectric, and desolvation free energy successfully discriminates the nativelike conformation of our submitted predictions. The latter emphasizes the critical role that thermodynamics plays on our methodology, and validates the generality of the algorithm to predict protein interactions.     

Performance of human and server prediction in CAPRI rounds 38-45

CAPRI challenges offer a variety of blind tests for protein-protein interaction prediction. In CAPRI Rounds 38-45, we generated a set of putative binding modes for each target with an FFT-based docking algorithm, and then scored and ranked these binding modes with a proprietary scoring function, ITScorePP. We have also developed a novel web server, Rebipp. The algorithm utilizes information retrieval to identify relevant biological information to significantly reduce the search space for a particular protein. In parallel, we have also constructed a GPU-based docking server, MDockPP, for protein-protein complex structure prediction. Here, the performance of our protocol in CAPRI rounds 38-45 is reported, which include 16 docking and scoring targets. Among them, three targets contain multiple interfaces: Targets 124, 125, and 136 have 2, 4, and 3 interfaces, respectively. In the predictor experiments, we predicted correct binding modes for nine targets, including one high-accuracy interface, six medium-accuracy binding modes, and six acceptable-accuracy binding modes. For the docking server prediction experiments, we predicted correct binding modes for eight targets, including one high-accuracy, three medium-accuracy, and five acceptable-accuracy binding modes.     

ClusPro: performance in CAPRI rounds 6-11 and the new server

ClusPro is the first fully automated, web-based program for docking protein structures. Users may upload the coordinate files of two protein structures through ClusPro's web interface, or enter the PDB codes of the respective structures. The server performs rigid body docking, energy screening, and clustering to produce models. The program output is a short list of putative complexes ranked according to their clustering properties. ClusPro has been participating in CAPRI since January 2003, submitting predictions within 24 h after a target becomes available. In Rounds 6-11, ClusPro generated acceptable submissions for Targets 22, 25, and 27. In general, acceptable models were obtained for the relatively easy targets without substantial conformational changes upon binding. We also describe the new version of ClusPro that incorporates our recently developed docking program PIPER. PIPER is based on the fast Fourier transform correlation approach, but the method is extended to use pairwise interaction potentials, thereby increasing the number of near-native docked structures.     

Docking proteins and peptides under evolutionary constraints in Critical Assessment of PRediction of Interactions rounds 38 to 45

Computational structural prediction of macromolecular interactions is a fundamental tool toward the global understanding of cellular processes. The Critical Assessment of PRediction of Interactions (CAPRI) community-wide experiment provides excellent opportunities for blind testing computational docking methods and includes original targets, thus widening the range of docking applications. Our participation in CAPRI rounds 38 to 45 enabled us to expand the way we include evolutionary information in structural predictions beyond our standard free docking InterEvDock pipeline. InterEvDock integrates a coarse-grained potential that accounts for interface coevolution based on joint multiple sequence alignments of two protein partners (co-alignments). However, even though such co-alignments could be built for none of the CAPRI targets in rounds 38 to 45, including host-pathogen and protein-oligosaccharide complexes and a redesigned interface, we identified multiple strategies that can be used to incorporate evolutionary constraints, which helped us to identify the most likely macromolecular binding modes. These strategies include template-based modeling where only local adjustments should be applied when query-template sequence identity is above 30% and larger perturbations are needed below this threshold; covariation-based structure prediction for individual protein partners; and the identification of evolutionarily conserved and structurally recurrent anchoring interface motifs. Overall, we submitted correct predictions among the top 5 models for 12 out of 19 interface challenges, including four High- and five Medium-quality predictions. Our top 20 models included correct predictions for three out of the five targets we missed in the top 5, including two targets for which misleading biological data led us to downgrade correct free docking models.     

Scoring a diverse set of high-quality docked conformations: a metascore based on electrostatic and desolvation interactions

Predicting protein-protein interactions involves sampling and scoring docked conformations. Barring some large structural rearrangement, rapidly sampling the space of docked conformations is now a real possibility, and the limiting step for the successful prediction of protein interactions is the scoring function used to reduce the space of conformations from billions to a few, and eventually one high affinity complex. An atomic level free-energy scoring function that estimates in units of kcal/mol both electrostatic and desolvation interactions (plus van der Waals if appropriate) of protein-protein docked conformations is used to rerank the blind predictions (860 in total) submitted for six targets to the community-wide Critical Assessment of PRediction of Interactions (CAPRI; http://capri.ebi.ac.uk). We found that native-like models often have varying intermolecular contacts and atom clashes, making unlikely that one can construct a universal function that would rank all these models as native-like. Nevertheless, our scoring function is able to consistently identify the native-like complexes as those with the lowest free energy for the individual models of 16 (out of 17) human predictors for five of the targets, while at the same time the modelers failed to do so in more than half of the cases. The scoring of high-quality models developed by a wide variety of methods and force fields confirms that electrostatic and desolvation forces are the dominant interactions determining the bound structure. The CAPRI experiment has shown that modelers can predict valuable models of protein-protein complexes, and improvements in scoring functions should soon solve the docking problem for complexes whose backbones do not change much upon binding. A scoring server and programs are available at http://structure.pitt.edu.     

Assessment of CAPRI predictions in rounds 3-5 shows progress in docking procedures

The current status of docking procedures for predicting protein-protein interactions starting from their three-dimensional (3D) structure is reassessed by evaluating blind predictions, performed during 2003-2004 as part of Rounds 3-5 of the community-wide experiment on Critical Assessment of PRedicted Interactions (CAPRI). Ten newly determined structures of protein-protein complexes were used as targets for these rounds. They comprised 2 enzyme-inhibitor complexes, 2 antigen-antibody complexes, 2 complexes involved in cellular signaling, 2 homo-oligomers, and a complex between 2 components of the bacterial cellulosome. For most targets, the predictors were given the experimental structures of 1 unbound and 1 bound component, with the latter in a random orientation. For some, the structure of the free component was derived from that of a related protein, requiring the use of homology modeling. In some of the targets, significant differences in conformation were displayed between the bound and unbound components, representing a major challenge for the docking procedures. For 1 target, predictions could not go to completion. In total, 1866 predictions submitted by 30 groups were evaluated. Over one-third of these groups applied completely novel docking algorithms and scoring functions, with several of them specifically addressing the challenge of dealing with side-chain and backbone flexibility. The quality of the predicted interactions was evaluated by comparison to the experimental structures of the targets, made available for the evaluation, using the well-agreed-upon criteria used previously. Twenty-four groups, which for the first time included an automatic Web server, produced predictions ranking from acceptable to highly accurate for all targets, including those where the structures of the bound and unbound forms differed substantially. These results and a brief survey of the methods used by participants of CAPRI Rounds 3-5 suggest that genuine progress in the performance of docking methods is being achieved, with CAPRI acting as the catalyst.     

Progress in protein-protein docking: atomic resolution predictions in the CAPRI experiment using RosettaDock with an improved treatment of side-chain flexibility

RosettaDock uses real-space Monte Carlo minimization (MCM) on both rigid-body and side-chain degrees of freedom to identify the lowest free energy docked arrangement of 2 protein structures. An improved version of the method that uses gradient-based minimization for off-rotamer side-chain optimization and includes information from unbound structures was used to create predictions for Rounds 4 and 5 of CAPRI. First, large numbers of independent MCM trajectories were carried out and the lowest free energy docked configurations identified. Second, new trajectories were started from these lowest energy structures to thoroughly sample the surrounding conformation space, and the lowest energy configurations were submitted as predictions. For all cases in which there were no significant backbone conformational changes, a small number of very low-energy configurations were identified in the first, global search and subsequently found to be close to the center of the basin of attraction in the free energy landscape in the second, local search. Following the release of the experimental coordinates, it was found that the centers of these free energy minima were remarkably close to the native structures in not only the rigid-body orientation but also the detailed conformations of the side-chains. Out of 8 targets, the lowest energy models had interface root-mean-square deviations (RMSDs) less than 1.1 A from the correct structures for 6 targets, and interface RMSDs less than 0.4 A for 3 targets. The predictions were top submissions to CAPRI for Targets 11, 12, 14, 15, and 19. The close correspondence of the lowest free energy structures found in our searches to the experimental structures suggests that our free energy function is a reasonable representation of the physical chemistry, and that the real space search with full side-chain flexibility to some extent solves the protein-protein docking problem in the absence of significant backbone conformational changes. On the other hand, the approach fails when there are significant backbone conformational changes as the steric complementarity of the 2 proteins cannot be modeled without incorporating backbone flexibility, and this is the major goal of our current work.     

Protein-protein docking using 3D-Dock in rounds 3, 4, and 5 of CAPRI

In rounds 3-5 of CAPRI, the community-wide experiment on the comparative evaluation of protein-protein docking for structure prediction, we applied the 3D-Dock software package to predict the atomic structures of nine biophysical interactions. This approach starts with an initial grid-based shape complementarity search. The product of this is a large number of potential interacting conformations that are subsequently ranked by interface residue propensities and interaction energies. Refinement through detailed energetics and optimization of side-chain positions using a rotamer library is also performed. For rounds 3, 4, and 5 of the CAPRI evaluation, where possible, we clustered functional residues on the surfaces of the monomers as an indication of binding sites, using sequence based evolutionary conservations. In certain targets this provided a very useful tool for identifying the areas of interaction. During round 5, we also applied the techniques of side-chain trimming and geometrical clustering described in the literature. Of the nine target complexes in rounds 3-5, we predicted conformations that contained at least some correct contact residues for seven of these systems. For two of the targets, we submitted predictions that were considered as medium-quality. These were a nidogen-laminin complex for target 8 (T08) and a serine-threonine phosphatase bound to a targeting subunit (T14). For a further three target systems, we produced models that were rated as acceptable predictions.     

ATTRACT: protein-protein docking in CAPRI using a reduced protein model

Protein-protein complex structures have been predicted for CAPRI Rounds 3 and 5 using a reduced protein model. Proteins are represented by up to 3 pseudoatoms per amino acid. The docking approach termed ATTRACT is based on energy minimization in translational and rotational degrees of freedom of one protein with respect to another protein. The reduced protein model allows one to perform systematic docking minimization of many thousand start structures in reasonable computer time. Flexibility of critical surface side-chains can be accounted for by a multiple conformational copy approach. The multicopy approach allows simultaneous adjustment of side-chain conformations and optimization of translational and rotational degrees of freedom of one protein with respect to the partner during docking. For 3 (Targets 8, 14, and 19) out of 5 CAPRI targets, the approach resulted in predictions in close agreement with experiment [root-mean-square deviation (RMSD) of backbone atoms within 10 A of the protein-protein interface < 1.8 A]. The comparison of predicted and experimental structures of the CAPRI targets indicates that besides local conformational changes (e.g., changes in side-chain conformations), global conformational changes of the protein backbone can be critical for complex formation. These conformational changes not accounted for during docking are a likely reason for the unrealistic predictions in 2 cases (Targets 9 and 18).     

CAPRI: a Critical Assessment of PRedicted Interactions

CAPRI is a communitywide experiment to assess the capacity of protein-docking methods to predict protein-protein interactions. Nineteen groups participated in rounds 1 and 2 of CAPRI and submitted blind structure predictions for seven protein-protein complexes based on the known structure of the component proteins. The predictions were compared to the unpublished X-ray structures of the complexes. We describe here the motivations for launching CAPRI, the rules that we applied to select targets and run the experiment, and some conclusions that can already be drawn. The results stress the need for new scoring functions and for methods handling the conformation changes that were observed in some of the target systems. CAPRI has already been a powerful drive for the community of computational biologists who development docking algorithms. We hope that this issue of Proteins will also be of interest to the community of structural biologists, which we call upon to provide new targets for future rounds of CAPRI, and to all molecular biologists who view protein-protein recognition as an essential process.     

Template-based modeling by ClusPro in CASP13 and the potential for using co-evolutionary information in docking

As a participant in the joint CASP13-CAPRI46 assessment, the ClusPro server debuted its new template-based modeling functionality. The addition of this feature, called ClusPro TBM, was motivated by the previous CASP-CAPRI assessments and by the proven ability of template-based methods to produce higher-quality models, provided templates are available. In prior assessments, ClusPro submissions consisted of models that were produced via free docking of pre-generated homology models. This method was successful in terms of the number of acceptable predictions across targets; however, analysis of results showed that purely template-based methods produced a substantially higher number of medium-quality models for targets for which there were good templates available. The addition of template-based modeling has expanded ClusPro's ability to produce higher accuracy predictions, primarily for homomeric but also for some heteromeric targets. Here we review the newest additions to the ClusPro web server and discuss examples of CASP-CAPRI targets that continue to drive further development. We also describe ongoing work not yet implemented in the server. This includes the development of methods to improve template-based models and the use of co-evolutionary information for data-assisted free docking.     

Assessment of blind predictions of protein-protein interactions: current status of docking methods

The current status of docking procedures for predicting protein-protein interactions starting from their three-dimensional structure is assessed from a first major evaluation of blind predictions. This evaluation was performed as part of a communitywide experiment on Critical Assessment of PRedicted Interactions (CAPRI). Seven newly determined structures of protein-protein complexes were available as targets for this experiment. These were the complexes between a kinase and its protein substrate, between a T-cell receptor beta-chain and a superantigen, and five antigen-antibody complexes. For each target, the predictors were given the experimental structures of the free components, or of one free and one bound component in a random orientation. The structure of the complex was revealed only at the time of the evaluation. A total of 465 predictions submitted by 19 groups were evaluated. These groups used a wide range of algorithms and scoring functions, some of which were completely novel. The quality of the predicted interactions was evaluated by comparing residue-residue contacts and interface residues to those in the X-ray structures and by analyzing the fit of the ligand molecules (the smaller of the two proteins in the complex) or of interface residues only, in the predicted versus target complexes. A total of 14 groups produced predictions, ranking from acceptable to highly accurate for five of the seven targets. The use of available biochemical and biological information, and in one instance structural information, played a key role in achieving this result. It was essential for identifying the native binding modes for the five correctly predicted targets, including the kinase-substrate complex where the enzyme changes conformation on association. But it was also the cause for missing the correct solution for the two remaining unpredicted targets, which involve unexpected antigen-antibody binding modes. Overall, this analysis reveals genuine progress in docking procedures but also illustrates the remaining serious limitations and points out the need for better scoring functions and more effective ways for handling conformational flexibility.     

The performance of ZDOCK and ZRANK in rounds 6-11 of CAPRI

We present an evaluation of our protein-protein docking approach using the ZDOCK and ZRANK algorithms, in combination with structural clustering and filtering, utilizing biological data in Rounds 6-11 of the CAPRI docking experiment. We achieved at least one prediction of acceptable accuracy for five of six targets submitted. In addition, two targets resulted in medium-accuracy predictions. In the new scoring portion of the CAPRI exercise, we were able to attain at least one acceptable prediction for the three targets submitted and achieved three medium-accuracy predictions for Target 26. Scoring was performed using ZRANK, a new algorithm for reranking initial-stage docking predictions using a weighted energy function and no structural refinement. Here we outline a practical and successful docking strategy, given limited prior biological knowledge of the complex to be predicted.     

The targets of CAPRI rounds 3-5

Ten protein-protein complexes have been offered by X-ray crystallographers as targets for structure prediction in Rounds 3-5 of the CAPRI experiment. They illustrate molecular recognition in several domains of biology: enzyme regulation, antigen-antibody recognition, signal transduction, and oligomer assembly. The targets presented various degrees of difficulty to the predictors, depending on their status (bound when components were taken from the complex, unbound when coming from independent structures of the free proteins), the amplitude of conformation changes, and the amount of biological information available. Predictors produced high-quality models of 6 of the targets, good models of 3 others, and failed only in 1 case, where the conformation change was particularly large. This result demonstrates significant progress relative to earlier rounds of CAPRI.     

Challenges and opportunities of automated protein-protein docking: HDOCK server vs human predictions in CAPRI Rounds 38-46

Protein-protein docking plays an important role in the computational prediction of the complex structure between two proteins. For years, a variety of docking algorithms have been developed, as witnessed by the critical assessment of prediction interactions (CAPRI) experiments. However, despite their successes, many docking algorithms often require a series of manual operations like modeling structures from sequences, incorporating biological information, and selecting final models. The difficulties in these manual steps have significantly limited the applications of protein-protein docking, as most of the users in the community are nonexperts in docking. Therefore, automated docking like a web server, which can give a comparable performance to human docking protocol, is pressingly needed. As such, we have participated in the blind CAPRI experiments for Rounds 38-45 and CASP13-CAPRI challenge for Round 46 with both our HDOCK automated docking web server and human docking protocol. It was shown that our HDOCK server achieved an “acceptable” or higher CAPRI-rated model in the top 10 submitted predictions for 65.5% and 59.1% of the targets in the docking experiments of CAPRI and CASP13-CAPRI, respectively, which are comparable to 66.7% and 54.5% for human docking protocol. Similar trends can also be observed in the scoring experiments. These results validated our HDOCK server as an efficient automated docking protocol for nonexpert users. Challenges and opportunities of automated docking are also discussed.     

RosettaDock in CAPRI rounds 6-12

A challenge in protein-protein docking is to account for the conformational changes in the monomers that occur upon binding. The RosettaDock method, which incorporates sidechain flexibility but keeps the backbone fixed, was found in previous CAPRI rounds (4 and 5) to generate docking models with atomic accuracy, provided that conformational changes were mainly restricted to protein sidechains. In the recent rounds of CAPRI (6-12), large backbone conformational changes occur upon binding for several target complexes. To address these challenges, we explicitly introduced backbone flexibility in our modeling procedures by combining rigid-body docking with protein structure prediction techniques such as modeling variable loops and building homology models. Encouragingly, using this approach we were able to correctly predict a significant backbone conformational change of an interface loop for Target 20 (12 A rmsd between those in the unbound monomer and complex structures), but accounting for backbone flexibility in protein-protein docking is still very challenging because of the significantly larger conformational space, which must be surveyed. Motivated by these CAPRI challenges, we have made progress in reformulating RosettaDock using a "fold-tree" representation, which provides a general framework for treating a wide variety of flexible-backbone docking problems.     

Prediction and scoring of docking poses with pyDock

The two previous CAPRI experiments showed the success of our rigid-body and refinement approach. For this third edition of CAPRI, we have used a new faster protocol called pyDock, which uses electrostatics and desolvation energy to score docking poses generated with FFT-based algorithms. In target T24 (unbound/model), our best prediction had the highest value of fraction of native contacts (40%) among all participants, although it was not considered as acceptable by the CAPRI criteria. In target T25 (unbound/bound), we submitted a model with medium quality. In target T26 (unbound/unbound), we did not submit any acceptable model (but we would have submitted acceptable predictions if we had included available mutational information about the binding site). For targets T27 (unbound/unbound) and T28 (homo-dimer using model), nobody (including us) submitted any acceptable model. Intriguingly, the crystal structure of target T27 shows an alternative interface that correlates with available biological data (we would have submitted acceptable predictions if we had included this). We also participated in all targets of the SCORERS experiment, with at least acceptable accuracy in all valid cases. We submitted two medium and four acceptable scoring models of T25. Using additional distance restraints (from mutational data), we had two medium and two acceptable scoring models of T26. For target T27, we submitted two acceptable scoring models of the alternative interface in the crystal structure. In summary, CAPRI showed the excellent capabilities of pyDock in identifying near-native docking poses.     

The targets of CAPRI rounds 6-12

Six protein-protein complexes and two homodimeric proteins involved in a variety of biological processes were offered as targets to CAPRI by crystallographers in Rounds 6-12. CAPRI predictor groups had to predict their structure by docking the free proteins, which they did with a degree of success that depended largely on the amplitude of the conformation changes. In one case at least, the prediction pointed to alternative possibilities of interactions in the crystal of a complex, showing that docking methods have value even when there is an experimental structure.