Conformational Ensembles from Experimental Data and Computer Simulations

91 / 161

Conformational Ensembles from Experimental Data
and Computer Simulations Poster Abstracts
87
52-POS Board 12
Fast High-Resolution Refinement of Large Proteins and Membrane Protein Complexes in
Rosetta
Julia Koehler Leman1,2, Jeffrey J. Gray3, Richard Bonneau1,2.
1Simons Foundation, New York, NY, USA, 2New York University, New York, NY, USA,
3Johns Hopkins University, Baltimore, MD, USA.
Protein structure refinement is one of the most often used modeling applications, especially for
high-resolution refinement of structures from homology modeling, de novo structure prediction,
loop modeling; to create starting models for protein design, protein-protein docking or ligand
docking; to refine structures into density maps from crystallography and cryoEM; to create all-
atom representations from coarse-grained models; and to model protein conformational
flexibility. Here, we present a new high-resolution refinement algorithm implemented into the
Rosetta software suite. Since Rosetta was historically developed for small, soluble proteins, the
gold-standard refinement algorithm in Rosetta, FastRelax, was neither routinely used, nor
benchmarked on large to very large, asymmetric proteins. When applied to larger complexes,
such as membrane proteins, FastRelax becomes prohibitively expensive. We show that our new
RangeRelax algorithm does not only speed up refinement of complexes up to ~5000 residues by
up to 60-fold, but it also results in fewer clashes and better stereochemistry. RangeRelax is
available for soluble and for membrane proteins and can be easily integrated into larger modeling
protocols.