PROTEIN STRUCTURE PREDICTION FROM NMR DATA

THIS PROJECT WILL DEVELOP A COMPUTER SOFTWARE PROGRAM THAT INCORPORATES EXPERIMENTAL NMR DATA INTO AUTOMATED PREDICTIONS OF THE THREE-DIMENSIONAL STRUCTURES OF PROTEINS. PHASE I WORK WILL INVOLVE THE DEVELOPMENT OF COMPUTER ALGORITHMS THAT RELIABLY AND EFFICIENTLY GENERATE PROTEIN STRUCTURES THAT AGREE WITH THE EXPERIMENTAL DATA. IN ADDITION TO CONFORMING TO EXPERIMENTAL DATA, CAPABILITIESWILL BE BUILT INTO THE ALGORITHMS TO ENSURE THAT THE PREDICTED STRUCTURES ARE REASONABLE IN TERMS OF THEIR DETAILED GEOMETRIC ANALYSIS AND THAT THE PREDICTED STRUCTURES CORRESPOND TO A STABLE ENERGY STRUCTURE FROM A MOLECULAR MECHANICS POINT OF VIEW. THE RESEARCH WILL INVESTIGATE NOE DISTANCE CONSTRAINT SCALING PROCEDURES, MONTE-CARLO TECHNIQUES TO AID IN THE SEARCHING OF CONFORMATIONAL SPACE, AND MANIPULATION OF ATOMIC SIZES TO GENERATE EFFICIENT STRUCTURE PREDICTION ALGORITHMS. DURING PHASE II, A COMMERCIAL SOFTWARE PACKAGE, FEATURING STATE-OF-THE-ART GRAPHICS AND MENU INTERFACES, WILL BE DEVELOPED TO ALLOW EASY USE AND IMPLEMENTATION OF THE ALGORITHMS DEVELOPED IN PHASE I. THE BASIC RESEARCH PERFORMED DURING THIS STUDY MAY LEAD TO SIGNIFICANT ADVANCESIN THE PREDICTION OF TERTIARY PROTEIN STRUCTURE SOLELY FROM A PROTEIN'S SEQUENCE.