Automatic model building &refinement in crystallography

European Molecular Biology Laboratory, Heidelberg

Investigators

Linked publications & trials

Abstract

DESCRIPTION (provided by applicant): The Protein Structure Initiative (Structural Genomics) aims to understand the structure and function of proteins and protein families, creating the knowledge that will need to be exploited for a wealth of biomedical applications. The main method to obtain the structures necessary for Structural Genomics is macromolecular X-ray crystallography. For Structural Genomics to succeed, it is important to possess methodology that allows rapid, large scale, automatic structure determination. The present research proposal aims to extend the ARP/wARP software for high-quality automated model building and refinement in macromolecular crystallography. This will lead to a major contribution towards automation and high-throughput which are necessary for the above projects and objectives. Traditionally, a research scientist has to build a molecular model into the experimentally available electron density map (the three-dimensional image of the molecule), a task often tedious, time demanding, subjective and heavily relying on experience. The initial macromolecular model then undergoes a refinement procedure in which the parameters of the model are adjusted to best fit the experimental data and stereochemical expectations. As the model improves, the electron density maps improve as well and a better model may be fitted. Model building and refinement are therefore related and tightly linked to each other, and should be regarded as one unified process. The ARP/wARP software, which we wish to further develop, fully automates the above procedure resulting in a faster, efficient, objective and reliable technique compared to traditional procedures. We plan to develop improved new algorithms for model re-parameterization and pattern recognition in three-dimensional space to improve the automated building of the initial macromolecular model. Genetic algorithms, Monte Carlo techniques and semi-definite programming will be exploited to aid in obtaining a more complete and accurate model. A graphical user interface and database and documentation service through the World Wide Web will help the structural biology community and more specifically the Structural Genomics centers to better exploit the developed software.