Use of Restraints from Consensus Fragments of Multiple Server Models To Enhance Protein-Structure Prediction Capability of the UNRES Force Field

Abstract

Recently, we developed a new approach to protein-structure prediction, which combines template-based modeling with the physics-based coarse-grained UNited RESidue (UNRES) force field. In this approach, restrained multiplexed replica exchange molecular dynamics simulations with UNRES, with the C-distance and virtual-bond-dihedral-angle restraints derived from knowledge-based models are carried out. In this work, we report a test of this approach in the 11th Community Wide Experiment on the Critical Assessment of Techniques for Protein Structure Prediction (CASP11), in which we used the template-based models from early-stage predictions by the LEE group CASP11 server (group 038, called "nns"), and further improvement of the method. The quality of the models obtained in CASP11 was better than that resulting from unrestrained UNRES simulations; however, the obtained models were generally worse than the final nns models. Calculations with the final nns models, performed after CASP11, resulted in substantial improvement, especially for multi-domain proteins. Based on these results, we modified the procedure by deriving restraints from models from multiple servers, in this study the four top-performing servers in CASP11 (nns, BAKER-ROSETTASERVER, Zhang-server, and QUARK), and implementing either all restraints or only the restraints on the fragments that appear similar in the majority of models (the consensus fragments), outlier models discarded. Tests with 29 CASP11 human-prediction targets with length less than 400 amino-acid residues demonstrated that the consensus-fragment approach gave better results, i.e., lower α-carbon root-mean-square deviation from the experimental structures, higher template modeling score, and global distance test total score values than the best of the parent server models. Apart from global improvement (repacking and improving the orientation of domains and other substructures), improvement was also reached for template-based modeling targets, indicating that the approach has refinement capacity. Therefore, the consensus-fragment analysis is able to remove lower-quality models and poor-quality parts of the models without knowing the experimental structure.

Ergodicity and model quality in template‐restrained canonical and temperature/Hamiltonian replica exchange coarse‐grained molecular dynamics simulations of proteins

“…The log‐Gaussian restraints on the distances, virtual‐bond angles, virtual‐bond‐dihedral angles, and local coordinates of side‐chain centers are expressed by eqs. , respectively; these terms originate from the MODELLER restraint pseudo‐energy function and are similar to those which were implemented in our earlier work …”

“…In our earlier work, standard deviations of the geometric parameters were assigned according to the confidence of secondary‐structure prediction for the respective residues or the confidence scores determined by the nns server, respectively. In this work, we calculated the standard deviations of the restraints from the accuracy of the TBM prediction of the geometry estimated by the QMEAN server, as expressed by eqs.…”

“…None of the server models used to derive the restraints captures the entire fold. In our previous study, using the consensus‐fragments from the nns, BAKER‐ROSETTASERVER, Zhang‐Server, and QUARK server models, we produced a model of better quality for T0828 but not for the remaining three proteins.…”

“…Recently, we proposed a hybrid approach to protein‐structure prediction, in which restrained conformational search is carried out with the use of the physics‐based coarse‐grained UNRES force field, subject to the distance‐ and angle‐restraints from knowledge‐based models. This approach is an extension of the application of the UNRES force field beyond unrestrained (physics‐based) simulations of protein structure and dynamics.…”

“…As opposed to model refinement with MODELLER, use of coarse‐grained representation of polypeptide chains enables us to carry out an extensive conformational search and restraints are imposed not on the whole structure but only on its parts that are predicted by the knowledge‐based methods with high confidence; these parts are domains or fragments for which most of the knowledge‐based methods predict a similar structure. We found that this approach often results in major improvements of the obtained models with respect to the parent models, because the restraints most compatible with the UNRES energy function are selected and, moreover, UNRES can build the part of structure which is undefined in the parent models . However, as all restrained simulations which aim at extensive search of the conformational space, the approach is potentially liable to ergodicity problems, because of the appearance of barriers due to the introduction of the restraints.…”