HIGH-PERFORMANCE DNA AND PROTEIN SEQUENCE ANALYSIS ON A LOW-COST SIMD PARALLEL PROCESSOR ARRAY

AS THE GROWTH OF DNA AND PROTEIN SEQUENCE DATABASES CONTINUES TO ACCELERATE, EXHAUSTIVE METHODS FOR SEARCHING THEM FOR HOMOLOGIES AND BIOLOGICALLY MEANINGFUL FEATURES HAVE BECOME INCREASINGLY IMPRACTICAL ON TRADITIONAL SERIAL-ARCHITECTURE COMPUTERS. RECENTLY, CONSIDERABLE WORK HAS BEEN DONE TO IMPLEMENT THESE METHODS ON SUPERCOMPUTERS AND HIGH-END MASSIVELY PARALLEL PROCESSORS, BUT THE EXPENSE AND POOR ACCESSIBILITY OF THESE MACHINES PLACES THEM OUT OF REACH OF MOST BIOLOGISTS. IN PHASE I, THIS PROJECT WILL FOCUS ON THE DEVELOPMENT AND IMPLEMENTATION OF EFFICIENT ALGORITHMS FOR THE COMPARISON AND EXHAUSTIVE DATABASE SEARCHING OF MACROMOLECULAR SEQUENCES ON A RELATIVELY INEXPENSIVE SIMD (SINGLE INSTRUCTION MULTIPLE DATA STREAM) PARALLEL ARRAY COMPUTER. VARIATIONS IN SCHEMES FOR SIMILARITY SCORING WILL BE EXAMINED, AND THE STATISTICAL PROPERTIES IF THESE SCORES IN THE CONTEXT OF DATABASE SEARCHES WILL BE EVALUATED. THROUGHOUT THE PROJECT, EMPHASIS WILL BE PLACED ON ACHIEVING HIGH PERFORMANCE IN A COST-EFFECTIVE MANNER. AN OBJECT-ORIENTED APPROACH TO SOFTWARE DESIGN AND DEVELOPMENT WILL BE PRACTICED FROM THE OUTSET TO MAXIMIZE EFFICIENCY DURING PHASES II AND III.