Automated x-ray crystallography

University Of Calif-Lawrenc Berkeley Lab, Berkeley CA

Investigators

Linked publications & trials

Abstract

This is project II of "Automated X-ray crystallography for Structural Genomics." We will extend the automation of structure solution we have implemented in our SOLVE software to a probability-based system applicable to all aspects of structure solution, density modification, model-building, and refinement. Additionally, we will develop the maximum-likelihood density modification procedure we have recently invented and to use it as a way to carry out all phase combination steps in crystal structure solution that involve calculations based on modified electron densities. These technologies will be an integral part of the Phenix automated structure-determination software, which in turns will be an essential tool for structural genomics and all allow rapid structure determination of proteins of medical, commercial, and academic interest. Decision-making. We will develop a simple decision-making framework that will allow automation of the entire structure determination process, from scaling to a final model. The key idea in our strategy is to formulate each decision in terms of its anticipated effects on two quantities. (1) a well-defined measure of quality such as the signal-to-noise in an electron density map, and (2) the time required to complete the process. This decision-making framework will be implemented in conjunction with projects I and IV. Maximum-likelihood density modification and phase recombination. We have recently developed maximum-likelihood density modification, a powerful approach for combination of real-space and reciprocal-space phase information. Our method provides for the first time a sound statistical basis for solvent flattening and other density-modification as an integrated part of this automated structure determination package. This part of the project will be closely tied to project II to develop probabilistic methods for identifying the solvent boundary in a reciprocal-space formulation. The maximum-likelihood density modification method will be extended to include non-crystallographic symmetry averaging, molecular replacement, and partial-model phasing. We will incorporate our recently-developed method for probabilistic pattern recognition of structural motifs in an electron density map and their use in maximum- likelihood density modification.