Ab Initio Prediction of 3-D Conformations for Protein Long Loops with High Accuracy and Applications to Antibody CDRH3 Modeling.

Residue-level potentials of mean force were widely used for protein backbone refinements to avoid simultaneous sampling of side-chain conformations. The interaction energy between the reduced side chains and backbone atoms was not considered explicitly. In this study, we developed novel methods to calculate the residue-atom interaction energy in combination with atomic and residue-level terms. The parameters were optimized step by step to remove the overcounting or overlap problem between different energy terms. The mixing energy functions were then used to evaluate the generated backbone conformations at the initial sampling stage of protein loop modeling (OSCAR-loop), including the interaction energy between the reduced loop residues and full atoms of the protein framework. The accuracies of top-ranked decoys were 1.18 and 2.81 Å for 8-residue and 12-residue loops, respectively. We then selected diverse decoys for side-chain modeling, backbone refinement, and energy minimization. The procedure was repeated multiple times to select one prediction with the lowest energy. Consequently, we obtained an accuracy of 0.74 Å for a prevailing test set of 12-residue loops, compared with >1.4 Å reported by other researchers. The OSCAR-loop was also effective for modeling the H3 loops of antibody complementary determining regions (CDRs) in the crystal environment. The prediction accuracy of OSCAR-loop (1.74 Å) was better than the accuracy of the Rosetta NGK method (3.11 Å) or those achieved by deep learning methods (>2.2 Å) for the CDRH3 loops of 49 targets in the Rosetta antibody benchmark. The performance of OSCAR-loop in a model environment was also discussed.

De novo protein design by an energy function based on series expansion in distance and orientation dependence.

MOTIVATION: Despite many successes, de novo protein design is not yet a solved problem as its success rate remains low. The low success rate is largely because we do not yet have an accurate energy function for describing the solvent-mediated interaction between amino acid residues in a protein chain. Previous studies showed that an energy function based on series expansions with its parameters optimized for side-chain and loop conformations can lead to one of the most accurate methods for side chain (OSCAR) and loop prediction (LEAP). Following the same strategy, we developed an energy function based on series expansions with the parameters optimized in four separate stages (recovering single-residue types without and with orientation dependence, selecting loop decoys and maintaining the composition of amino acids). We tested the energy function for de novo design by using Monte Carlo simulated annealing. RESULTS: The method for protein design (OSCAR-Design) is found to be as accurate as OSCAR and LEAP for side-chain and loop prediction, respectively. In de novo design, it can recover native residue types ranging from 38% to 43% depending on test sets, conserve hydrophobic/hydrophilic residues at ∼75%, and yield the overall similarity in amino acid compositions at more than 90%. These performance measures are all statistically significantly better than several protein design programs compared. Moreover, the largest hydrophobic patch areas in designed proteins are near or smaller than those in native proteins. Thus, an energy function based on series expansion can be made useful for protein design. AVAILABILITY AND IMPLEMENTATION: The Linux executable version is freely available for academic users at http://zhouyq-lab.szbl.ac.cn/resources/.

Kotai Antibody Builder: automated high-resolution structural modeling of antibodies.

MOTIVATION: Kotai Antibody Builder is a Web service for tertiary structural modeling of antibody variable regions. It consists of three main steps: hybrid template selection by sequence alignment and canonical rules, 3D rendering of alignments and CDR-H3 loop modeling. For the last step, in addition to rule-based heuristics used to build the initial model, a refinement option is available that uses fragment assembly followed by knowledge-based scoring. Using targets from the Second Antibody Modeling Assessment, we demonstrate that Kotai Antibody Builder generates models with an overall accuracy equal to that of the best-performing semi-automated predictors using expert knowledge. AVAILABILITY AND IMPLEMENTATION: Kotai Antibody Builder is available at http://kotaiab.org CONTACT: standley@ifrec.osaka-u.ac.jp.

High-resolution modeling of antibody structures by a combination of bioinformatics, expert knowledge, and molecular simulations.

In the second antibody modeling assessment, we used a semiautomated template-based structure modeling approach for 11 blinded antibody variable region (Fv) targets. The structural modeling method involved several steps, including template selection for framework and canonical structures of complementary determining regions (CDRs), homology modeling, energy minimization, and expert inspection. The submitted models for Fv modeling in Stage 1 had the lowest average backbone root mean square deviation (RMSD) (1.06 Å). Comparison to crystal structures showed the most accurate Fv models were generated for 4 out of 11 targets. We found that the successful modeling in Stage 1 mainly was due to expert-guided template selection for CDRs, especially for CDR-H3, based on our previously proposed empirical method (H3-rules) and the use of position specific scoring matrix-based scoring. Loop refinement using fragment assembly and multicanonical molecular dynamics (McMD) was applied to CDR-H3 loop modeling in Stage 2. Fragment assembly and McMD produced putative structural ensembles with low free energy values that were scored based on the OSCAR all-atom force field and conformation density in principal component analysis space, respectively, as well as the degree of consensus between the two sampling methods. The quality of 8 out of 10 targets improved as compared with Stage 1. For 4 out of 10 Stage-2 targets, our method generated top-scoring models with RMSD values of less than 1 Å. In this article, we discuss the strengths and weaknesses of our approach as well as possible directions for improvement to generate better predictions in the future.

LEAP: highly accurate prediction of protein loop conformations by integrating coarse-grained sampling and optimized energy scores with all-atom refinement of backbone and side chains.

Prediction of protein loop conformations without any prior knowledge (ab initio prediction) is an unsolved problem. Its solution will significantly impact protein homology and template-based modeling as well as ab initio protein-structure prediction. Here, we developed a coarse-grained, optimized scoring function for initial sampling and ranking of loop decoys. The resulting decoys are then further optimized in backbone and side-chain conformations and ranked by all-atom energy scoring functions. The final integrated technique called loop prediction by energy-assisted protocol achieved a median value of 2.1 Å root mean square deviation (RMSD) for 325 12-residue test loops and 2.0 Å RMSD for 45 12-residue loops from critical assessment of structure-prediction techniques (CASP) 10 target proteins with native core structures (backbone and side chains). If all side-chain conformations in protein cores were predicted in the absence of the target loop, loop-prediction accuracy only reduces slightly (0.2 Å difference in RMSD for 12-residue loops in the CASP target proteins). The accuracy obtained is about 1 Å RMSD or more improvement over other methods we tested. The executable file for a Linux system is freely available for academic users at http://sparks-lab.org.

Community-wide evaluation of methods for predicting the effect of mutations on protein-protein interactions.

Community-wide blind prediction experiments such as CAPRI and CASP provide an objective measure of the current state of predictive methodology. Here we describe a community-wide assessment of methods to predict the effects of mutations on protein-protein interactions. Twenty-two groups predicted the effects of comprehensive saturation mutagenesis for two designed influenza hemagglutinin binders and the results were compared with experimental yeast display enrichment data obtained using deep sequencing. The most successful methods explicitly considered the effects of mutation on monomer stability in addition to binding affinity, carried out explicit side-chain sampling and backbone relaxation, evaluated packing, electrostatic, and solvation effects, and correctly identified around a third of the beneficial mutations. Much room for improvement remains for even the best techniques, and large-scale fitness landscapes should continue to provide an excellent test bed for continued evaluation of both existing and new prediction methodologies.

PITHA: A Webtool to Predict Immunogenicity for Humanized and Fully Human Therapeutic Antibodies.

Immunogenicity is an important concern to therapeutic antibodies during antibody design and development. Based on the co-crystal structures of idiotypic antibodies and their antibodies, one can see that anti-idiotypic antibodies usually bind the complementarity-determining regions (CDR) of idiotypic antibodies. Sequence and structural features, such as cavity volume at the CDR region and hydrophobicity of CDR-H3 loop region, were identified for distinguishing immunogenic antibodies from non-immunogenic antibodies. These features were integrated together with a machine learning platform to predict immunogenicity for humanized and fully human therapeutic antibodies (PITHA). This method achieved an accuracy of 83% in a leave-one-out experiment for 29 therapeutic antibodies with available crystal structures. The web server of this method is accessible at http://mabmedicine.com/PITHA or http://sysbio.unl.edu/PITHA . This method, as a step of computer-aided antibody design, helps evaluate the safety of new therapeutic antibody, which can save time and money during the therapeutic antibody development.

B-Cell Epitope Predictions Using Computational Methods.

Identifying protein antigenic epitopes that are recognizable by antibodies is a key step in immunologic research. This type of research has broad medical applications, such as new immunodiagnostic reagent discovery, vaccine design, and antibody design. However, due to the countless possibilities of potential epitopes, the experimental search through trial and error would be too costly and time-consuming to be practical. To facilitate this process and improve its efficiency, computational methods were developed to predict both linear epitopes and discontinuous antigenic epitopes. For linear B-cell epitope prediction, many methods were developed, including PREDITOP, PEOPLE, BEPITOPE, BepiPred, COBEpro, ABCpred, AAP, BCPred, BayesB, BEOracle/BROracle, BEST, LBEEP, DRREP, iBCE-EL, SVMTriP, etc. For the more challenging yet important task of discontinuous epitope prediction, methods were also developed, including CEP, DiscoTope, PEPITO, ElliPro, SEPPA, EPITOPIA, PEASE, EpiPred, SEPIa, EPCES, EPSVR, etc. In this chapter, we will discuss computational methods for B-cell epitope predictions of both linear and discontinuous epitopes. SVMTriP and EPCES/EPCSVR, the most successful among the methods for each type of the predictions, will be used as model methods to detail the standard protocols. For linear epitope prediction, SVMTriP was reported to achieve a sensitivity of 80.1% and a precision of 55.2% with a fivefold cross-validation based on a large dataset, yielding an AUC of 0.702. For discontinuous or conformational B-cell epitope prediction, EPCES and EPCSVR were both benchmarked by a curated independent test dataset in which all antigens had no complex structures with the antibody. The identified epitopes by these methods were later independently validated by various biochemical experiments. For these three model methods, webservers and all datasets are publicly available at http://sysbio.unl.edu/SVMTriP , http://sysbio.unl.edu/EPCES/ , and http://sysbio.unl.edu/EPSVR/ .

Fast and accurate prediction of protein side-chain conformations.

SUMMARY: We developed a fast and accurate side-chain modeling program [Optimized Side Chain Atomic eneRgy (OSCAR)-star] based on orientation-dependent energy functions and a rigid rotamer model. The average computing time was 18 s per protein for 218 test proteins with higher prediction accuracy (1.1% increase for χ(1) and 0.8% increase for χ(1+2)) than the best performing program developed by other groups. We show that the energy functions, which were calibrated to tolerate the discrete errors of rigid rotamers, are appropriate for protein loop selection, especially for decoys without extensive structural refinement. AVAILABILITY: OSCAR-star and the 218 test proteins are available for download at http://sysimm.ifrec.osaka-u.ac.jp/OSCAR CONTACT: standley@ifrec.osaka-u.ac.jp SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Protein loop selection using orientation-dependent force fields derived by parameter optimization.

We used the orientation-dependent Optimized Side Chain Atomic eneRgy (OSCAR-o), derived in an early study, for protein loop selection. The prediction accuracy of OSCAR-o was better than that of physics-based force fields or statistical potential energy functions for both the RAPPER decoy set and the Jacobson decoy set. The native conformer was frequently ranked as lowest energy among the decoys. Furthermore, strong correlation was observed between the OSCAR-o score and the root mean square deviation (RMSD) from the native structure for energy-minimized decoys. In practical use, we applied OSCAR-o to rescore decoys generated by a widely used loop-modeling program, LOOPY. As a result, the mean RMSD values of top-ranked decoys were reduced by 0.3 Å for loop targets of seven to nine residues. We expect similar performance for OSCAR-o with other loop-modeling algorithms in the context of decoy rescoring. A loop selection program (OSCAR-ls) based on OSCAR-o is available at http://sysimm.ifrec.osaka-u.ac.jp/OSCAR/.

Protein side chain modeling with orientation-dependent atomic force fields derived by series expansions.

We describe the development of new force fields for protein side chain modeling called optimized side chain atomic energy (OSCAR). The distance-dependent energy functions (OSCAR-d) and side-chain dihedral angle potential energy functions were represented as power and Fourier series, respectively. The resulting 802 adjustable parameters were optimized by discriminating the native side chain conformations from non-native conformations, using a training set of 12,000 side chains for each residue type. In the course of optimization, for every residue, its side chain was replaced by varying rotamers, whereas conformations for all other residues were kept as they appeared in the crystal structure. Then, the OSCAR-d were multiplied by an orientation-dependent function to yield OSCAR-o. A total of 1087 parameters of the orientation-dependent energy functions (OSCAR-o) were optimized by maximizing the energy gap between the native conformation and subrotamers calculated as low energy by OSCAR-d. When OSCAR-o with optimized parameters were used to model side chain conformations simultaneously for 218 recently released protein structures, the prediction accuracies were 88.8% for χ(1) , 79.7% for χ(1 + 2) , 1.24 Å overall root mean square deviation (RMSD), and 0.62 Å RMSD for core residues, respectively, compared with the next-best performing side-chain modeling program which achieved 86.6% for χ(1) , 75.7% for χ(1 + 2) , 1.40 Å overall RMSD, and 0.86 Å RMSD for core residues, respectively. The continuous energy functions obtained in this study are suitable for gradient-based optimization techniques for protein structure refinement. A program with built-in OSCAR for protein side chain prediction is available for download at http://sysimm.ifrec.osaka-u.ac.jp/OSCAR/.

EPSVR and EPMeta: prediction of antigenic epitopes using support vector regression and multiple server results.

BACKGROUND: Accurate prediction of antigenic epitopes is important for immunologic research and medical applications, but it is still an open problem in bioinformatics. The case for discontinuous epitopes is even worse - currently there are only a few discontinuous epitope prediction servers available, though discontinuous peptides constitute the majority of all B-cell antigenic epitopes. The small number of structures for antigen-antibody complexes limits the development of reliable discontinuous epitope prediction methods and an unbiased benchmark to evaluate developed methods. RESULTS: In this work, we present two novel server applications for discontinuous epitope prediction: EPSVR and EPMeta, where EPMeta is a meta server. EPSVR, EPMeta, and datasets are available at http://sysbio.unl.edu/services. CONCLUSION: The server application for discontinuous epitope prediction, EPSVR, uses a Support Vector Regression (SVR) method to integrate six scoring terms. Furthermore, we combined EPSVR with five existing epitope prediction servers to construct EPMeta. All methods were benchmarked by our curated independent test set, in which all antigens had no complex structures with the antibody, and their epitopes were identified by various biochemical experiments. The area under the receiver operating characteristic curve (AUC) of EPSVR was 0.597, higher than that of any other existing single server, and EPMeta had a better performance than any single server - with an AUC of 0.638, significantly higher than PEPITO and Disctope (p-value < 0.05).

Prediction of immunogenicity for humanized and full human therapeutic antibodies.

Immunogenicity is an important concern for therapeutic antibodies during drug development. By analyzing co-crystal structures of idiotypic antibodies and their antibodies, we found that anti-idiotypic antibodies usually bind the Complementarity Determining Regions (CDR) of idiotypic antibodies. Sequence and structural features were identified for distinguishing immunogenic antibodies from non-immunogenic antibodies. For example, non-immunogenic antibodies have a significantly larger cavity volume at the CDR region and a more hydrophobic CDR-H3 loop than immunogenic antibodies. Antibodies containing no Gly at the turn of CDR-H2 loop are often immunogenic. We integrated these features together with a machine learning platform to Predict Immunogenicity for humanized and full human THerapeutic Antibodies (PITHA). This method achieved an accuracy of 83% in leave-one-out experiment for 29 therapeutic antibodies with available crystal structures. The accuracy decreased to 65% for 23 test antibodies with modeled structures, because their crystal structures were not available, and the prediction was made with modeled structures. The server of this method is accessible at http://mabmedicine.com/PITHA.

EPCES and EPSVR: Prediction of B-Cell Antigenic Epitopes on Protein Surfaces with Conformational Information.

Accurate prediction of discontinuous antigenic epitopes is important for immunologic research and medical applications, but it is not an easy problem. Currently, there are only a few prediction servers available, though discontinuous epitopes constitute the majority of all B-cell antigenic epitopes. In this chapter, we describe two online servers, EPCES and EPSVR, for discontinuous epitope prediction. All methods were benchmarked by a curated independent test set, in which all antigens had no complex structures with the antibody, and their epitopes were identified by various biochemical experiments. The servers and all datasets are available at http://sysbio.unl.edu/EPCES/ and http://sysbio.unl.edu/EPSVR/ .

SVMTriP: A Method to Predict B-Cell Linear Antigenic Epitopes.

Identifying protein antigenic epitopes recognizable by antibodies is the key step for new immuno-diagnostic reagent discovery and vaccine design. To facilitate this process and improve its efficiency, computational methods were developed to predict antigenic epitopes. For the linear B-cell epitope prediction, many methods were developed, including BepiPred, ABCPred, AAP, BCPred, BayesB, BEOracle/BROracle, BEST, and SVMTriP. Among these methods, SVMTriP, a frontrunner, utilized Support Vector Machine by combining the tri-peptide similarity and Propensity scores. Applied on non-redundant B-cell linear epitopes extracted from IEDB, SVMTriP achieved a sensitivity of 80.1% and a precision of 55.2% with a five-fold cross-validation. The AUC value was 0.702. The combination of similarity and propensity of tri-peptide subsequences can improve the prediction performance for linear B-cell epitopes. A webserver based on this method was constructed for public use. The server and all datasets used in the corresponding study are available at http://sysbio.unl.edu/SVMTriP . This chapter describes the webserver of SVMTriP.

Prediction of antigenic epitopes on protein surfaces by consensus scoring.

BACKGROUND: Prediction of antigenic epitopes on protein surfaces is important for vaccine design. Most existing epitope prediction methods focus on protein sequences to predict continuous epitopes linear in sequence. Only a few structure-based epitope prediction algorithms are available and they have not yet shown satisfying performance. RESULTS: We present a new antigen Epitope Prediction method, which uses ConsEnsus Scoring (EPCES) from six different scoring functions - residue epitope propensity, conservation score, side-chain energy score, contact number, surface planarity score, and secondary structure composition. Applied to unbounded antigen structures from an independent test set, EPCES was able to predict antigenic eptitopes with 47.8% sensitivity, 69.5% specificity and an AUC value of 0.632. The performance of the method is statistically similar to other published methods. The AUC value of EPCES is slightly higher compared to the best results of existing algorithms by about 0.034. CONCLUSION: Our work shows consensus scoring of multiple features has a better performance than any single term. The successful prediction is also due to the new score of residue epitope propensity based on atomic solvent accessibility.

Exploring the molecular design of protein interaction sites with molecular dynamics simulations and free energy calculations.

The significant work that has been invested toward understanding protein-protein interaction has not translated into significant advances in structure-based predictions. In particular redesigning protein surfaces to bind to unrelated receptors remains a challenge, partly due to receptor flexibility, which is often neglected in these efforts. In this work, we computationally graft the binding epitope of various small proteins obtained from the RCSB database to bind to barnase, lysozyme, and trypsin using a previously derived and validated algorithm. In an effort to probe the protein complexes in a realistic environment, all native and designer complexes were subjected to a total of nearly 400 ns of explicit-solvent molecular dynamics (MD) simulation. The MD data led to an unexpected observation: some of the designer complexes were highly unstable and decomposed during the trajectories. In contrast, the native and a number of designer complexes remained consistently stable. The unstable conformers provided us with a unique opportunity to define the structural and energetic factors that lead to unproductive protein-protein complexes. To that end we used free energy calculations following the MM-PBSA approach to determine the role of nonpolar effects, electrostatics and entropy in binding. Remarkably, we found that a majority of unstable complexes exhibited more favorable electrostatics than native or stable designer complexes, suggesting that favorable electrostatic interactions are not prerequisite for complex formation between proteins. However, nonpolar effects remained consistently more favorable in native and stable designer complexes reinforcing the importance of hydrophobic effects in protein-protein binding. While entropy systematically opposed binding in all cases, there was no observed trend in the entropy difference between native and designer complexes. A series of alanine scanning mutations of hot-spot residues at the interface of native and designer complexes showed less than optimal contacts of hot-spot residues with their surroundings in the unstable conformers, resulting in more favorable entropy for these complexes. Finally, disorder predictions revealed that secondary structures at the interface of unstable complexes exhibited greater disorder than the stable complexes.

Refining near-native protein-protein docking decoys by local resampling and energy minimization.

How to refine a near-native structure to make it closer to its native conformation is an unsolved problem in protein-structure and protein-protein complex-structure prediction. In this article, we first test several scoring functions for selecting locally resampled near-native protein-protein docking conformations and then propose a computationally efficient protocol for structure refinement via local resampling and energy minimization. The proposed method employs a statistical energy function based on a Distance-scaled Ideal-gas REference state (DFIRE) as an initial filter and an empirical energy function EMPIRE (EMpirical Protein-InteRaction Energy) for optimization and re-ranking. Significant improvement of final top-1 ranked structures over initial near-native structures is observed in the ZDOCK 2.3 decoy set for Benchmark 1.0 (74% whose global rmsd reduced by 0.5 A or more and only 7% increased by 0.5 A or more). Less significant improvement is observed for Benchmark 2.0 (38% versus 33%). Possible reasons are discussed.

Consensus scoring for enriching near-native structures from protein-protein docking decoys.

The identification of near native protein-protein complexes among a set of decoys remains highly challenging. A strategy for improving the success rate of near native detection is to enrich near native docking decoys in a small number of top ranked decoys. Recently, we found that a combination of three scoring functions (energy, conservation, and interface propensity) can predict the location of binding interface regions with reasonable accuracy. Here, these three scoring functions are modified and combined into a consensus scoring function called ENDES for enriching near native docking decoys. We found that all individual scores result in enrichment for the majority of 28 targets in ZDOCK2.3 decoy set and the 22 targets in Benchmark 2.0. Among the three scores, the interface propensity score yields the highest enrichment in both sets of protein complexes. When these scores are combined into the ENDES consensus score, a significant increase in enrichment of near-native structures is found. For example, when 2000 dock decoys are reduced to 200 decoys by ENDES, the fraction of near-native structures in docking decoys increases by a factor of about six in average. ENDES was implemented into a computer program that is available for download at http://sparks.informatics.iupui.edu.

Protein binding site prediction using an empirical scoring function.

Most biological processes are mediated by interactions between proteins and their interacting partners including proteins, nucleic acids and small molecules. This work establishes a method called PINUP for binding site prediction of monomeric proteins. With only two weight parameters to optimize, PINUP produces not only 42.2% coverage of actual interfaces (percentage of correctly predicted interface residues in actual interface residues) but also 44.5% accuracy in predicted interfaces (percentage of correctly predicted interface residues in the predicted interface residues) in a cross validation using a 57-protein dataset. By comparison, the expected accuracy via random prediction (percentage of actual interface residues in surface residues) is only 15%. The binding sites of the 57-protein set are found to be easier to predict than that of an independent test set of 68 proteins. The average coverage and accuracy for this independent test set are 30.5 and 29.4%, respectively. The significant gain of PINUP over expected random prediction is attributed to (i) effective residue-energy score and accessible-surface-area-dependent interface-propensity, (ii) isolation of functional constraints contained in the conservation score from the structural constraints through the combination of residue-energy score (for structural constraints) and conservation score and (iii) a consensus region built on top-ranked initial patches.