Piet Gros, Dept. of Crystal and Structural Chemistry, Bijvoet Center for Biomolecular Research, Utrecht University, Padualaan 18, 3584 CH Utrecht, The Netherlands (gros@chem.ruu.nl)
A method is presented for ab initio phasing of protein and peptide data sets at medium resolution; examples of phasing peptide data sets up to low resolution (ca. 4.5 Ang) will be presented. The optimization method consists primarily of solvent flattening and structure refinement (using Molecular Dynamics and Energy Minimization). To make structure refinement feasible for random starting models a force field is introduced that is applicable to loose and all equal atoms. Pseudo-energy potential functions (potentials of mean force) in this force field are derived from radial-distribution functions of all non-hydrogen atoms as is observed in refined structures. Thus, the configuration of atoms is restrained towards atomic distributions of protein structures. Solvent flattening is applied to estimate the bulk solvent contribution to the structure factors. This procedure has been tested for a few peptide data sets. The data was generously supplied by Dr Isabella Karle. In the tests the data were truncated to 2.5 Ang resolution, because data to atomic resolution is rarely available when a protein structure is being determined. The resulting models from these optimization show a correspondence at low resolution to the known structures. Analysis of the phase differences and map correlation coefficients (w.r.t. the known answers) indicates that phase information up to approx. 4.5 Ang is obtained.