High-Throughput cDNA Screening Utilizing a Low Order Neural Network Filter. (December 1996)

• Record Type: Journal Article
• Authors: Huang, Guyang Matthew; Farkas, James; Hood, Leroy
• Is Part Of: Biotechniques. Volume 21:Number 6(1996)
• Page Start: 1110
• Page End: 1114
• Publication Date: 1996-12
• Subjects: Biology, Experimental -- Periodicals; Molecular biology -- Periodicals; Medical technology -- Periodicals; Biology, Experimental; Medical technology; Molecular biology; Clinical Laboratory Techniques -- Periodicals; Research -- Periodicals; Medical Laboratory Science -- Periodicals; Periodicals; Electronic journals; 570
• DOI: 10.2144/96216bc01
• Languages: English
• ISSNs: 0736-6205

Abstract

A low order neural network-based filter was designed as a rapid screening agent for single-spanning transmembrane regions in an integrated informatics system. A rapid screening algorithm was seen as a compromise between costly structure-specific techniques and simple rules that gave a high false-positive rate for cDNA. The filter was applied to a library of 2123 anonymous cDNA sequences, which resulted in 61 detections. Evaluation of the detections with two other dissimilar computer prediction algorithms yielded strong transmembrane predictions for 15 of the detections, while 8 of the detections resulted in a definitive negative result. Homology searches performed on the sequences with detection reports yielded 13 homologs in the predicted reading frame, four of which are membrane associated.