Technology/Business Opportunity

Non-Confidential Description

University of California - Santa Cruz Campus

MULTI-PURPOSE PARALLEL PROCESSOR FOR BIOSEQUENCE ANALYSIS
DESCRIPTION: Biologists store DNA, RNA, and protein sequences in large databases that are then used to compare and classify sequences. These databases are growing so quickly that single processor computers are no longer adequate to support efficient searches or analysis of sequences. Current parallel machines and systems have their drawbacks. Supercomputers are expensive and slower than hardware implementations of algorithms. Coprocessors offer better performance and cost but are either limited to one algorithm or, if reconfigurable, can require large amounts of reprogramming time and an in depth knowledge of the hardware. Additionally, since existing coprocessors only perform sequence comparison, not sequence alignment, they can not provide information about the correspondence between parts of two sequences.

Scientists at the University of California have designed a system with massively parallel VLSI architecture particularly suited to the dynamic programming used in sequence analysis applications. This system not only supports standard biosequence search and analysis, but also may be applied to sequence alignment and experimental algorithms. It is designed to be implemented with scalable CMOS technology and fabricated by MOSIS.

APPLICATIONS: The system can support standard biosequence analysis methods such as Smith and Waterman, profile, BLAST, and dictionary searches. Additionally, the system supports sequence alignment and hidden Markov model training and searching. The system will also support algorithms developed in the future.

ADVANTAGES:

INQUIRIES TO: Kathy S. Willis, Licensing Associate
Tel: (510) 748-6600
Fax: (510) 748-6639
Email: kathy.willis@ucop.edu

REFERENCE: UC Case 95-154