An analysis and evaluation of the WeFold collaborative for protein structure prediction and its pipelines in CASP11 and CASP12.

Every two years groups worldwide participate in the Critical Assessment of Protein Structure Prediction (CASP) experiment to blindly test the strengths and weaknesses of their computational methods. CASP has significantly advanced the field but many hurdles still remain, which may require new ideas and collaborations. In 2012 a web-based effort called WeFold, was initiated to promote collaboration within the CASP community and attract researchers from other fields to contribute new ideas to CASP. Members of the WeFold coopetition (cooperation and competition) participated in CASP as individual teams, but also shared components of their methods to create hybrid pipelines and actively contributed to this effort. We assert that the scale and diversity of integrative prediction pipelines could not have been achieved by any individual lab or even by any collaboration among a few partners. The models contributed by the participating groups and generated by the pipelines are publicly available at the WeFold website providing a wealth of data that remains to be tapped. Here, we analyze the results of the 2014 and 2016 pipelines showing improvements according to the CASP assessment as well as areas that require further adjustments and research.

Princeton_TIGRESS 2.0: High refinement consistency and net gains through support vector machines and molecular dynamics in double-blind predictions during the CASP11 experiment.

Protein structure refinement is the challenging problem of operating on any protein structure prediction to improve its accuracy with respect to the native structure in a blind fashion. Although many approaches have been developed and tested during the last four CASP experiments, a majority of the methods continue to degrade models rather than improve them. Princeton_TIGRESS (Khoury et al., Proteins 2014;82:794-814) was developed previously and utilizes separate sampling and selection stages involving Monte Carlo and molecular dynamics simulations and classification using an SVM predictor. The initial implementation was shown to consistently refine protein structures 76% of the time in our own internal benchmarking on CASP 7-10 targets. In this work, we improved the sampling and selection stages and tested the method in blind predictions during CASP11. We added a decomposition of physics-based and hybrid energy functions, as well as a coordinate-free representation of the protein structure through distance-binning Cα-Cα distances to capture fine-grained movements. We performed parameter estimation to optimize the adjustable SVM parameters to maximize precision while balancing sensitivity and specificity across all cross-validated data sets, finding enrichment in our ability to select models from the populations of similar decoys generated for targets in CASPs 7-10. The MD stage was enhanced such that larger structures could be further refined. Among refinement methods that are currently implemented as web-servers, Princeton_TIGRESS 2.0 demonstrated the most consistent and most substantial net refinement in blind predictions during CASP11. The enhanced refinement protocol Princeton_TIGRESS 2.0 is freely available as a web server at http://atlas.engr.tamu.edu/refinement/. Proteins 2017; 85:1078-1098. © 2017 Wiley Periodicals, Inc.

conSSert: Consensus SVM Model for Accurate Prediction of Ordered Secondary Structure.

Accurate prediction of protein secondary structure remains a crucial step in most approaches to the protein-folding problem, yet the prediction of ordered secondary structure, specifically beta-strands, remains a challenge. We developed a consensus secondary structure prediction method, conSSert, which is based on support vector machines (SVM) and provides exceptional accuracy for the prediction of beta-strands with QE accuracy of over 0.82 and a Q2-EH of 0.86. conSSert uses as input probabilities for the three types of secondary structure (helix, strand, and coil) that are predicted by four top performing methods: PSSpred, PSIPRED, SPINE-X, and RAPTOR. conSSert was trained/tested using 4261 protein chains from PDBSelect25, and 8632 chains from PISCES. Further validation was performed using targets from CASP9, CASP10, and CASP11. Our data suggest that poor performance in strand prediction is likely a result of training bias and not solely due to the nonlocal nature of beta-sheet contacts. conSSert is freely available for noncommercial use as a webservice: http://ares.tamu.edu/conSSert/.

New compstatin peptides containing N-terminal extensions and non-natural amino acids exhibit potent complement inhibition and improved solubility characteristics.

Compstatin peptides are complement inhibitors that bind and inhibit cleavage of complement C3. Peptide binding is enhanced by hydrophobic interactions; however, poor solubility promotes aggregation in aqueous environments. We have designed new compstatin peptides derived from the W4A9 sequence (Ac-ICVWQDWGAHRCT-NH2, cyclized between C2 and C12), based on structural, computational, and experimental studies. Furthermore, we developed and utilized a computational framework for the design of peptides containing non-natural amino acids. These new compstatin peptides contain polar N-terminal extensions and non-natural amino acid substitutions at positions 4 and 9. Peptides with α-modified non-natural alanine analogs at position 9, as well as peptides containing only N-terminal polar extensions, exhibited similar activity compared to W4A9, as quantified via ELISA, hemolytic, and cell-based assays, and showed improved solubility, as measured by UV absorbance and reverse-phase HPLC experiments. Because of their potency and solubility, these peptides are promising candidates for therapeutic development in numerous complement-mediated diseases.

De novo design and experimental characterization of ultrashort self-associating peptides.

Self-association is a common phenomenon in biology and one that can have positive and negative impacts, from the construction of the architectural cytoskeleton of cells to the formation of fibrils in amyloid diseases. Understanding the nature and mechanisms of self-association is important for modulating these systems and in creating biologically-inspired materials. Here, we present a two-stage de novo peptide design framework that can generate novel self-associating peptide systems. The first stage uses a simulated multimeric template structure as input into the optimization-based Sequence Selection to generate low potential energy sequences. The second stage is a computational validation procedure that calculates Fold Specificity and/or Approximate Association Affinity (K*association) based on metrics that we have devised for multimeric systems. This framework was applied to the design of self-associating tripeptides using the known self-associating tripeptide, Ac-IVD, as a structural template. Six computationally predicted tripeptides (Ac-LVE, Ac-YYD, Ac-LLE, Ac-YLD, Ac-MYD, Ac-VIE) were chosen for experimental validation in order to illustrate the self-association outcomes predicted by the three metrics. Self-association and electron microscopy studies revealed that Ac-LLE formed bead-like microstructures, Ac-LVE and Ac-YYD formed fibrillar aggregates, Ac-VIE and Ac-MYD formed hydrogels, and Ac-YLD crystallized under ambient conditions. An X-ray crystallographic study was carried out on a single crystal of Ac-YLD, which revealed that each molecule adopts a ß-strand conformation that stack together to form parallel ß-sheets. As an additional validation of the approach, the hydrogel-forming sequences of Ac-MYD and Ac-VIE were shuffled. The shuffled sequences were computationally predicted to have lower K*association values and were experimentally verified to not form hydrogels. This illustrates the robustness of the framework in predicting self-associating tripeptides. We expect that this enhanced multimeric de novo peptide design framework will find future application in creating novel self-associating peptides based on unnatural amino acids, and inhibitor peptides of detrimental self-aggregating biological proteins.

Forcefield_NCAA: ab initio charge parameters to aid in the discovery and design of therapeutic proteins and peptides with unnatural amino acids and their application to complement inhibitors of the compstatin family.

We describe the development and testing of ab initio derived, AMBER ff03 compatible charge parameters for a large library of 147 noncanonical amino acids including ß- and N-methylated amino acids for use in applications such as protein structure prediction and de novo protein design. The charge parameter derivation was performed using the RESP fitting approach. Studies were performed assessing the suitability of the derived charge parameters in discriminating the activity/inactivity between 63 analogs of the complement inhibitor Compstatin on the basis of previously published experimental IC50 data and a screening procedure involving short simulations and binding free energy calculations. We found that both the approximate binding affinity (K*) and the binding free energy calculated through MM-GBSA are capable of discriminating between active and inactive Compstatin analogs, with MM-GBSA performing significantly better. Key interactions between the most potent Compstatin analog that contains a noncanonical amino acid are presented and compared to the most potent analog containing only natural amino acids and native Compstatin. We make the derived parameters and an associated web interface that is capable of performing modifications on proteins using Forcefield_NCAA and outputting AMBER-ready topology and parameter files freely available for academic use at http://selene.princeton.edu/FFNCAA . The forcefield allows one to incorporate these customized amino acids into design applications with control over size, van der Waals, and electrostatic interactions.

WeFold: a coopetition for protein structure prediction.

The protein structure prediction problem continues to elude scientists. Despite the introduction of many methods, only modest gains were made over the last decade for certain classes of prediction targets. To address this challenge, a social-media based worldwide collaborative effort, named WeFold, was undertaken by 13 labs. During the collaboration, the laboratories were simultaneously competing with each other. Here, we present the first attempt at "coopetition" in scientific research applied to the protein structure prediction and refinement problems. The coopetition was possible by allowing the participating labs to contribute different components of their protein structure prediction pipelines and create new hybrid pipelines that they tested during CASP10. This manuscript describes both successes and areas needing improvement as identified throughout the first WeFold experiment and discusses the efforts that are underway to advance this initiative. A footprint of all contributions and structures are publicly accessible at http://www.wefold.org.

Princeton_TIGRESS: protein geometry refinement using simulations and support vector machines.

Protein structure refinement aims to perform a set of operations given a predicted structure to improve model quality and accuracy with respect to the native in a blind fashion. Despite the numerous computational approaches to the protein refinement problem reported in the previous three CASPs, an overwhelming majority of methods degrade models rather than improve them. We initially developed a method tested using blind predictions during CASP10 which was officially ranked in 5th place among all methods in the refinement category. Here, we present Princeton_TIGRESS, which when benchmarked on all CASP 7,8,9, and 10 refinement targets, simultaneously increased GDT_TS 76% of the time with an average improvement of 0.83 GDT_TS points per structure. The method was additionally benchmarked on models produced by top performing three-dimensional structure prediction servers during CASP10. The robustness of the Princeton_TIGRESS protocol was also tested for different random seeds. We make the Princeton_TIGRESS refinement protocol freely available as a web server at http://atlas.princeton.edu/refinement. Using this protocol, one can consistently refine a prediction to help bridge the gap between a predicted structure and the actual native structure.

Protein folding and de novo protein design for biotechnological applications.

In the postgenomic era, the medical/biological fields are advancing faster than ever. However, before the power of full-genome sequencing can be fully realized, the connection between amino acid sequence and protein structure, known as the protein folding problem, needs to be elucidated. The protein folding problem remains elusive, with significant difficulties still arising when modeling amino acid sequences lacking an identifiable template. Understanding protein folding will allow for unforeseen advances in protein design; often referred to as the inverse protein folding problem. Despite challenges in protein folding, de novo protein design has recently demonstrated significant success via computational techniques. We review advances and challenges in protein structure prediction and de novo protein design, and highlight their interplay in successful biotechnological applications.

Protein WISDOM: a workbench for in silico de novo design of biomolecules.

The aim of de novo protein design is to find the amino acid sequences that will fold into a desired 3-dimensional structure with improvements in specific properties, such as binding affinity, agonist or antagonist behavior, or stability, relative to the native sequence. Protein design lies at the center of current advances drug design and discovery. Not only does protein design provide predictions for potentially useful drug targets, but it also enhances our understanding of the protein folding process and protein-protein interactions. Experimental methods such as directed evolution have shown success in protein design. However, such methods are restricted by the limited sequence space that can be searched tractably. In contrast, computational design strategies allow for the screening of a much larger set of sequences covering a wide variety of properties and functionality. We have developed a range of computational de novo protein design methods capable of tackling several important areas of protein design. These include the design of monomeric proteins for increased stability and complexes for increased binding affinity. To disseminate these methods for broader use we present Protein WISDOM (http://www.proteinwisdom.org), a tool that provides automated methods for a variety of protein design problems. Structural templates are submitted to initialize the design process. The first stage of design is an optimization sequence selection stage that aims at improving stability through minimization of potential energy in the sequence space. Selected sequences are then run through a fold specificity stage and a binding affinity stage. A rank-ordered list of the sequences for each step of the process, along with relevant designed structures, provides the user with a comprehensive quantitative assessment of the design. Here we provide the details of each design method, as well as several notable experimental successes attained through the use of the methods.

Forcefield_PTM: Ab Initio Charge and AMBER Forcefield Parameters for Frequently Occurring Post-Translational Modifications.

In this work, we introduce Forcefield_PTM, a set of AMBER forcefield parameters consistent with ff03 for 32 common post-translational modifications. Partial charges were calculated through ab initio calculations and a two-stage RESP-fitting procedure in an ether-like implicit solvent environment. The charges were found to be generally consistent with others previously reported for phosphorylated amino acids, and trimethyllysine, using different parameterization methods. Pairs of modified and their corresponding unmodified structures were curated from the PDB for both single and multiple modifications. Background structural similarity was assessed in the context of secondary and tertiary structures from the global dataset. Next, the charges derived for Forcefield_PTM were tested on a macroscopic scale using unrestrained all-atom Langevin molecular dynamics simulations in AMBER for 34 (17 pairs of modified/unmodified) systems in implicit solvent. Assessment was performed in the context of secondary structure preservation, stability in energies, and correlations between the modified and unmodified structure trajectories on the aggregate. As an illustration of their utility, the parameters were used to compare the structural stability of the phosphorylated and dephosphorylated forms of OdhI. Microscopic comparisons between quantum and AMBER single point energies along key χ torsions on several PTMs were performed and corrections to improve their agreement in terms of mean squared errors and squared correlation coefficients were parameterized. This forcefield for post-translational modifications in condensed-phase simulations can be applied to a number of biologically relevant and timely applications including protein structure prediction, protein and peptide design, docking, and to study the effect of PTMs on folding and dynamics. We make the derived parameters and an associated interactive webtool capable of performing post-translational modifications on proteins using Forcefield_PTM available at http://selene.princeton.edu/FFPTM.

Proteome-wide post-translational modification statistics: frequency analysis and curation of the swiss-prot database.

Post-translational modifications (PTMs) broadly contribute to the recent explosion of proteomic data and possess a complexity surpassing that of protein design. PTMs are the chemical modification of a protein after its translation, and have wide effects broadening its range of functionality. Based on previous estimates, it is widely believed that more than half of proteins are glycoproteins. Whereas mutations can only occur once per position, different forms of post-translational modifications may occur in tandem. With the number and abundances of modifications constantly being discovered, there is no method to readily assess their relative levels. Here we report the relative abundances of each PTM found experimentally and putatively, from high-quality, manually curated, proteome-wide data, and show that at best, less than one-fifth of proteins are glycosylated. We make available to the academic community a continuously updated resource (http://selene.princeton.edu/PTMCuration) containing the statistics so scientists can assess "how many" of each PTM exists.

Computational design of Candida boidinii xylose reductase for altered cofactor specificity.

In this study we introduce a computationally-driven enzyme redesign workflow for altering cofactor specificity from NADPH to NADH. By compiling and comparing data from previous studies involving cofactor switching mutations, we show that their effect cannot be explained as straightforward changes in volume, hydrophobicity, charge, or BLOSUM62 scores of the residues populating the cofactor binding site. Instead, we find that the use of a detailed cofactor binding energy approximation is needed to adequately capture the relative affinity towards different cofactors. The implicit solvation models Generalized Born with molecular volume integration and Generalized Born with simple switching were integrated in the iterative protein redesign and optimization (IPRO) framework to drive the redesign of Candida boidinii xylose reductase (CbXR) to function using the non-native cofactor NADH. We identified 10 variants, out of the 8,000 possible combinations of mutations, that improve the computationally assessed binding affinity for NADH by introducing mutations in the CbXR binding pocket. Experimental testing revealed that seven out of ten possessed significant xylose reductase activity utilizing NADH, with the best experimental design (CbXR-GGD) being 27-fold more active on NADH. The NADPH-dependent activity for eight out of ten predicted designs was either completely abolished or significantly diminished by at least 90%, yielding a greater than 10(4)-fold change in specificity to NADH (CbXR-REG). The remaining two variants (CbXR-RTT and CBXR-EQR) had dual cofactor specificity for both nicotinamide cofactors.