Pre-filtering improves reliability of Affymetrix GeneChips results when used to analyze gene expression in complex tissues

Affymetrix GeneChip represents a very reliable and standardized technology for genome-wide gene expression screening. However, in experiments carried out on complex biological samples (e.g. brain tissues composed of several diverse cell types), significant noise can arise due to important transcripts being expressed in a relatively small number of cells. This noise results in many observations coming from unreliable hybridization reactions. Here we propose a method for pre-filtering Affymetrix data according to measures of hybridization reliability. We used our pre-filtering method on a microarray dataset obtained from the brains of rats chronically treated with a psychostimulant drug. Our pre-filter protocol facilitates selection of biologically relevant candidate genes, which could be validated by real-time PCR with a rate of 98%.