Study Weekend Program...
Bijvoet Center for Biomolecular Research
Utrecht University, Netherlands
In automated modeling procedures like Shake'N'Bake, ARP/wARP and RESOLVE, phasing and modeling are becoming more and more intertwined. At lower resolutions, and with poor or no (!) external phase information, modeling increasingly relies on a proper estimation of the quality of the intermediate fragmentary models obtained. A procedure that combines a loose atom modeling approach with molecular-dynamics type of optimization, called Conditional Optimization will be discussed. This procedure combines a continuous expression of the geometric quality of fragmentary models with a potentially large radius of convergence for optimization of loose atoms. We will show results obtained when attempting ab initio phasing using 2 Å data, and when applying this method to automated model-building tasks using low to medium resolution data (2.4-3.0 Å) with good experimental phases.