Abstract # A-25

Fish Species Identification Using Computer-Assisted Pattern Recognition with Protein Patterns from the World Wide Web- Accessible Regulatory Fish Encyclopedia (RFE) and with Remote Pattern Matching Capability. B. Tenge1, N. Dang1, F. Fry2, S. Bloch3, W. Savary4, P. Rogers2, J. Barnett1, W. Hill1, and M. Wekell1, 1Seafood Products Res. Ctr., & SEA-DO, FDA, Bothell, WA, 2CFSAN, FDA, Wash., D.C., 3Infometrix, Inc., Woodinville, WA, 4SAN-DO, FDA, Alameda, CA.

Isoelectric focusing (IEF) protein gel banding pattern analysis is used to identify a fish species when only a fillet is available to determine a case of economic fraud. IEF protein patterns can be digitized into densitograms and peak reports and then formatted according to the RFE (http://vm.cfsan.fda.gov/~frf/rfe0.html) gel-lane-layout format which includes specimen patterns co-migrated with standard pI markers for internal migration position normalization. Normalized peak reports are numerically regrouped ("binned") according to peak position into a set of overlapping bins (standardized variables), each representing a specific pI range. Thus, each fish specimen pattern is transformed into its chemometric pattern vector for SIMCA pattern recognition modeling. Specimen patterns from the cod, snapper, and rockfish families are acquired and modeled with SIMCA class distances > 2. Prediction of fish species identity using this system of SIMCA models has been enabled for use by remote Internet/Intranet users through WWW CGI.