ISB Home



- Article -





Volume 4


Full article

In Silico Biology 4, 0020 (2004); ©2004, Bioinformation Systems e.V.  



Global gene expression analysis by combinatorial optimization

Adam Ameur, Erik Aurell1, Mats Carlsson* and Jakub Orzechowski Westholm*

SICS, Swedish Institute of Computer Science, P.O. Box 1263, S-164 29 Kista, Sweden
Email: matsc@sics.se

1 Present address: KTH Royal Institute of Technology, AlbaNova University Center, S-106 91 Stockholm, Sweden

*  corresponding author


Edited by H. Michael; received October 30, 2003; revised and accepted February 25, 2004; published March 21, 2004


Abstract

Generally, there is a trade-off between methods of gene expression analysis that are precise but labor-intensive, e.g. RT-PCR, and methods that scale up to global coverage but are not quite as quantitative, e.g. microarrays. In the present paper, we show how how a known method of gene expression profiling (K. Kato, Nucleic Acids Res. 23, 3685-3690 (1995)), which relies on a fairly small number of steps, can be turned into a global gene expression measurement by advanced data post-processing, with potentially little loss of accuracy. Post-processing here entails solving an ancillary combinatorial optimization problem. Validation is performed on in silico experiments generated from the FANTOM data base of full-length mouse cDNA. We present two variants of the method. One uses state-of-the-art commercial software for solving problems of this kind, the other a code developed by us specifically for this purpose, released in the public domain under GPL license.

Key words: global gene expression, combinatorial optimization