Identification of Chromosomal Regions Containing Transcribed Sequences Using Microarray Expression Data

• Activity Number: 61
• Type: Contributed
• Date/Time: Sunday, August 3, 2003 : 4:00 PM to 5:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Lisa H. Ying*+ and Eric Schadt and Vladimir B. Svetnik and Daniel J. Holder and Stephen Edwards and Debraj GuhaThakurta
• Companies: Merck & Co., Inc. and Rosetta Inpharmatics and Merck and Company and Merck Research Laboratories and Merck & Co., Inc. and Merck & Co., Inc.
• Keywords: mMicroarray ; transcribed sequences ; principal component analysis ; clustering ; genome sequencing and annotation

Abstract

Current genome sequencing and annotation effects have produced a reasonably well annotated version of the human genome. However, a complete characterization of the human transcriptome still remains to be completed. We have developed novel methods to identify transcribed regions in genomic sequence from microarray-based hybridization patterns. Specifically, custom ink jet microarrays, manufactured by Agilent Technologies, Inc. of Palo Alto, CA, were designed with reporters that spanned two human chromosomes. RNA from eight tissue samples were hybridized to the arrays, and the data were processed using a two-step procedure to identify regions transcribed in at least one of the samples. First, we split the data into overlapping 15 kb windows and used robust PCA to identify windows likely to contain transcribed sequences. Then, for regions identified in the first step, we used clustering methods to discriminate between reporters corresponding to the transcribed and untranscribed regions. Our method achieves reasonably good sensitivity while maintaining a low false positive rate and results in identification of novel transcribed sequences and alternative forms of well-characterized genes.