The rapid growth in size of DNA and protein databases that will accompany the human genome project is sure to overburden available computer resources. Homology searches, optimal sequence comparisons, and studies of RNA secondary structure are all important molecular biology research topics, requiring the availability of powerful database and computational facilities. This investigation will explore the relationships between popular and emerging computerized sequence analysis algorithms, including DNA/ protein sequence mapping and RNA folding programs, and their fundamental computer architectural requirements. A baseline system, configured from commercially available elements, will be analyzed. If implemented, it would provide 100 million instructions per second of equivalent processing power to networked research teams at the equivalent cost of a superminicomputer.Anticipated Results/Potential Commercial Applications as described by the awardee: This processing system should provide high levels of computing power for all applications involving non-numerical string searches and analysis. In particular, large textual databases will be efficiently and rapidly accessed, and the database accelerator CPU will be architecturally configured to provide for symbolic sequence comparisons using dynamic-programming based algorithms. This type of processing is valuable in many database applications where soft information must be compared and related, using reasoning as the basis for decision making.Topic 4: Human Genome
