Table of Contents Author Guidelines Submit a Manuscript
Comparative and Functional Genomics
Volume 4, Issue 2, Pages 171-181
Primary Research

Gene Selection in Arthritis Classification with Large-Scale Microarray Expression Profiles

1Mathematical Sciences Department, University of Texas at El Paso, El Paso, TX 79968-0514, USA
2Department of Statistics, Texas A&M University, 3143 TAMU, College Station, TX 77843-3143, USA
3Institute of Mathematics and Statistics, University of Kent at Canterbury, Kent, Canterbury CT2 7NF, UK
4Genomics and Proteomic Sciences Division, GlaxoSmithKline, Medicines Research Centre, Stevenage, UK
5Statistical Sciences, GlaxoSmithKline, Medicines Research Centre, Stevenage, UK
6School of Biosciences, University of Birmingham, Birmingham B15 2TT, UK

Received 2 August 2002; Revised 18 January 2003; Accepted 30 January 2003

Copyright © 2003 Hindawi Publishing Corporation. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


The use of large-scale microarray expression profiling to identify predictors of disease class has become of major interest. Beyond their impact in the clinical setting (i.e. improving diagnosis and treatment), these markers are also likely to provide clues on the molecular mechanisms underlining the diseases. In this paper we describe a new method for the identification of multiple gene predictors of disease class. The method is applied to the classification of two forms of arthritis that have a similar clinical endpoint but different underlying molecular mechanisms: rheumatoid arthritis (RA) and osteoarthritis (OA). We aim at both the classification of samples and the location of genes characterizing the different classes. We achieve both goals simultaneously by combining a binary probit model for classification with Bayesian variable selection methods to identify important genes.We find very small sets of genes that lead to good classification results. Some of the selected genes are clearly correlated with known aspects of the biology of arthritis and, in some cases, reflect already known differences between RA and OA.