Automated x-ray crystallography

University Of Calif-Lawrenc Berkeley Lab, Berkeley CA

Investigators

Linked publications & trials

Abstract

Our long-term goal in this collaborative project is to develop maximum likelihood applications for protein crystallography, within the computational framework provided by PHENIX. By providing a better statistical basis for crystallographic calculations., likelihood increases the power for optimization methods. This increases in power will increase the success rate of the automated procedures required for high throughput structure determination in the context of the structural genomics. Applications will be developed for t he following crystallographic procedures: Structure refinement. The existing likelihood targets (MLF, MLI and MLHL), along with new likelihood targets based on joint probability distributions of structure factors from related structures, will be implemented within PHENIX.. The new targets will allow simultaneous phasing and refinement, and show be more effective in evaluating the true differences between related structures. Molecular replacement. With slight modifications, the refinement likelihood target is an excellent score function for molecular replacement computations, more powerful than traditional Patterson-based scores. Experimental phasing. Likelihood targets that allow for either uncorrelated lack-of-isomorphism errors or correlated lack- of isomorphism errors (important for multi-wavelength) anomalous diffraction phasing) will be implemented, as will a likelihood-based approach to determining the heavy-atom substructure. Density modification. A reciprocal-space likelihood function will allow the automatic evaluation of optimal envelopes for solvent flattening and averaging, as well as providing a valid statistical basis for evaluating the probability distributions of modified phases.