Year 1 - A pilot study on database generation and knowledge mining on the all-a class of protein will be undertaken. This study will identify most of the problems facing the entire project. The rules mined for the all-a protein will be evaluated.
Year 2 - Improvements (as dentified in year 1) to the knowledge mining algorithm will be made. Then the a+b class of proteins will be added to test the procedure on a larger more complex dataset to identify any scaling problems. The topological rules found will be assessed.
Year 3 - Final version of learning algorithm will be run on the entire database of protein structures. The topological rules will be evaluated. Rules will be mined to probe the relationship between protein function and structure. Development of a Web server to disseminate database, rules and explore development of a facility to process new protein coordinates. Final evaluation and dissemination of results.