Kouichi NITTA^* and Jun TANIDA¹

Graduate School of Engineering, Osaka University, 2-1 Yamadaoka, Suita, Osaka 565-0871, Japan
¹Graduate School of Information Science and Technology, Osaka University, 2-1 Yamadaoka, Suita, Osaka 565-0871, Japan

(Received August 30, 2002; Accepted November 25, 2002)

In this paper, we propose a novel inference method for a gene network using optical array logic (OAL).^1-3) OAL is utilized as a descriptive language for multipurpose parallel information processing on general electronic hardware. We develop an effective algorithm to extract direct regulations in a redundant gene network provided by biological experiments. In the proposed method, a gene interaction matrix is converted into a two-dimensional discrete binary image. The image is processed using parallel operations provided by OAL. After several steps of OAL operations indirect regulations in the gene network can be eliminated. The algorithm is believed to avoid undesired elimination of links caused by loop structures. In the simulation using artificial network data, the proposed technique achieves more than 90% sensitivity.

Key words: optical array logic, inference of gene network, bioinformatics, parallel processing, SIMD operation

^*E-mail address: nitta@photonics.mls.eng.osaka-u.ac.jp