Selection bias in gene extraction in tumour classification
by
Geoff McLachlan
University of Queensland
Coauthors: Christophe Ambroise (University of Technology Compiegne)

We describe some recent work of Ambroise and McLachlan (Proceedings of the National Academy of Sciences USA, Vol. 99, 2002) on the selection bias in gene extraction in tumour classification on the basis of microarray gene expression data. The problem concerns the construction of an accurate prediction rule on the basis of a relatively small number of tumour tissue samples of known type, containing the expression data on very many (possibly thousands) of genes. Recently, results have been presented in the literature, suggesting that it is possible to construct a prediction rule from only a small number of genes so that it has a negligible prediction error rate. In this talk, we examine the validity of these claims.

Date received: April 3, 2002