Source:From Computational Biophysics to Systems Biology (CBSB11) Proceedings, 8:29–32, 2012
We have designed a new pipeline for protein structure prediction based on the CABS engine. The procedure is fully automated and generates consensus models from a set of templates. Restraints derived from the templates define a region of conformational space, which is then sampled by Replica Exchange Monte Carlo algorithm implemented in CABS. Results from CASP9 show, that for great majority of targets this approach leads to better models than the mean quality of templates (in respect to GDT TS). In five cases the obtained models were the best among all predictions submitted to CASP9 as the first models.