Open Collections

Structure prediction of mono- and oligomeric tarets using the physics-based coarse-grained UNRES force field and information from databases â CASP13/CAPRI46

Banff International Research Station for Mathematical Innovation and Discovery

The results of blind prediction of the structures of monomeric and oligomeric proteins obtained in the recent CASP/CAPRI experiment by the KIAS-Gdansk/Czaplewski groups, by using the physics-based coarse-grained UNRES force field [1] and information from databases are presented. For both monomeric and oligomeric targets, the methodology of the KIAS-Gdansk/Czaplewski groups included extensive conformational search by means of the Multiplexed Replica Exchange Molecular Dynamics (MREMD) simulations [2] with the UNRES force field [3], with geometry restraints from server models [4,5]. For the monomeric targets, the restraints were derived from fragments with similar geometry that occured in the server models (the consensus fragments), while monomer geometries except for the terminal, flexible loop, and linker regions, were restrained in oligomer simulations. The server models were selected mainly based on DeepQA score [6] ranking; when the score was below 0.5, models from the servers which performed well in previous CASP exercises: Zhang, Quark, and BAKER-ROSETTASERVER were selected. For oligomeric targers, monomers were modeled first and, subsequently, initial oligomeric structures were modeled based with the aid of the packing proposed by the HHpred server [7]; for smaller targets the monomers were oriented randomly. The results of MREMD simulations were processed by using the Weighted Histogram Analysis Method (WHAM) [8] to obtain the probabilities of conformations and subsequently subjected to cluster analysis to obtain the 5 (CASP) or 10 (CAPRI) families of conformations, from which the conformations closest to the mean conformations were, in turn, selected as candidate predictions [1], which were subsequently converted to all-atom conformations submitted to CASP/CAPRI. The obtained models were ranked solely based on the computed probabilities of the families obtained by summing up the probabilities of the constituent conformations computed by WHAM based on the UNRES effective function [1].

Acknowledgments: This research was supported by grant UMO-2017/26/M/ST4/00044 from the National Science Centre of Poland (Narodowe Centrum Nauki).

[1] A. Liwo et al., J. Mol. Model. 20 (2014): 2306.
[2] Y.M. Rhee et al., Biophys. J. 84 (2003): 775-786.
[3] C. Czaplewski et al., J. Chem. Theor. Comput. 5 (2009): 627-640.
[4] P. Krupa et al., J. Chem. Inf. Model. 55 (2015): 1271-1281.
[5] M. Mozolewska et al., J. Chem. Inf. Model. 56 (2016): 2263-2279.
[6] R. Cao et al., BMC Bioinformatics. 17 (2016): 495. <rb> [7] L. Zimmermann et al., J Mol Biol. 430 (2018): 2237-2243.
[8] S. Kumar et al., J. Comput. Chem., (2001) 8, 1011-1021.

Mathematics; Biology and other natural sciences; Statistical mechanics, structure of matter; Mathematical biology

video/mp4

Author affiliation: University of Gdansk

2020-02-07

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/73494

Unreviewed

http://creativecommons.org/licenses/by-nc-nd/4.0/