EVOLUTIONARY ALGORITHM FOR PROTEIN STRUCTURE PREDICTION

The goal of this project is to develop a computer algorithm that accurately predicts the tertiary fold of a small protein given only its amino acid sequence and secondary structure. It will use an evolutionary optimization method with proprietary enhancements invented by the PI. The specific aims of Phase I are to: (1) develop a more natural representation of the problem; (2) implement an energy function appropriate for this representation; (3) improve the optimization algorithm; and (4) combine these into a simulator that successfully predicts tertiary from secondary structure. In Phase II this simulator will be extended to predict the tertiary structures of larger proteins from their sequences and from sparse experimental constraints or imperfect secondary structure predictions. It will be validated using many proteins of different sizes and fold classes and developed into software products for the pharmaceutical and biotechnology industries. The long-term goal is to help satisfy the rapidly growing demand for the accurate prediction of protein structures from sequence information. This demand arises not only from the exponential growth of the database of sequences, but also from the need to understand the structure and function of newly discovered gene products known to be involved in human disease.