Unified automated model building & refinement for biological crystallography

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Structural Biology in general and the Protein Structure Initiative in particular, aim to understand the structure and function of macromolecules, their complexes and families, creating a knowledge that is further explored for a wealth of biomedical applications. X-ray crystallography has become the most commonly used method to assist the investigation into biological phenomena by providing detailed atomic models of the bio- molecules of interest. To maximize the efficiency of X-ray crystallography, automation of labor intensive, repetitive tasks in protein structure determination is crucial. As such, the step of building an atomic model in the electron density map has to be made fast, reliable and highly automated. The ARP/wARP software has pioneered this automation step and helped to obtain a large number of novel structures of macromolecules. Scientific developments in ARP/wARP, mostly funded by the NIH over the last three years, landmarked considerable advancement of the overall software package. The developed algorithms and scientific concepts allow construction of more complete models and to extend the interpretation of lower resolution electron density maps. The software became easier to use for non-expert researchers via graphical user interfaces and WWW-based execution of remotely submitted tasks. The overall performance of the ARP/wARP software in terms of speed and convergence was improved. In the course of the requested renewal of the grant we will extend the aims of the project towards seamless automation of macromolecular 3-D structure determination, while we will deliver more complete and validated models with lower resolution of the experimental diffraction data. We will achieve our goals by developing further the pattern recognition-based algorithms; improving interlinks between different steps of structure determination with emphasis to large and multimeric structures; delivering truly complete models including poorly ordered surface regions, with special emphasis to building bound ligands; developing an 'expert control system' that would be capable of basic decision making based on the accumulated history; and by continuing to improve the accessibility of the software by the community.