PROJECT 1 - LAWRENCE BERKELEY LAB - PHENIX

点击翻译按钮获取中文摘要

Project I: During the first 5 years of the PHENIX project we have implemented a program for automated analysis of crystallographic data (mmtbx.xtriage), we have developed a highly automated program for anomalous substructure location (HySS), and we have created a new automated structure refinement program (phenix.refine). We have played an essential role in the birth of PHENIX by developing the open source C++ Computational Crystallography Toolbox (cctbx) and the PHENIX graphical user interface (GUI). We have worked closely with the other members of the PHENIX project to create an integrated system for automated structure determination that provides command-line interfaces, a visual programming interface, and the highly automated Wizard interface. We will use our strong foundation of computational algorithms and software to address some of the most challenging bottlenecks remaining in structure determination; classification of experimental data for decisionmaking, substructure location from weak signals, and productive and minimally-biased structure refinement at all resolution ranges. Our new algorithms which automatically assign a probability score to each dataset that describes the likelihood of successful structure solution will be used in subsequent decision-making by Projects II and IV. The result will be an increase in the efficiency of structure solution. The chances of successful substructure location will be improved by new algorithms for automated space group determination, optimal data cutoff calculation, and the application of improved likelihood scoring functions, developed in collaboration with Project III. This unique capability will make it possible for PHENIX to successfully-begin the process of structure solution, where other systems currently fail. We will develop new algorithms, including the use of normal models, for the refinement of models against experimental data at any resolution limit, by using automatic model parameterizations that maintain a reasonable ratio of parameters to observed data. We will also extend structure refinement to include algorithms for the local rebuilding of models, including peptide flips and rotamer refitting, in collaboration with Projects II and V. This approach will speed the convergence of refinement and reduce the need for manual intervention in many cases. Finally, we will enhance the PHENIX software by further development of our cctbx library, the PHENIX GUI, Project Data Storage, and the developer environment. Our new algorithms will enable researchers to determine the structures of challenging, high-value biological macromolecules which are important for understanding biology and ultimately improving human health.