Quality measures for Affymetrix data

Ken Simpson1
1ksimpson@wehi.edu.au, The Walter and Eliza Hall Institute

We present several methods (qualitative and quantitative) for determining the quality of hybridizations to Affymetrix GeneChips. We show how fitting the data robustly can identify and downweight low quality probes, as well as highlighting any spatial artifacts. An attempt is made to quantify the effect of such artifacts on the estimated gene expression, using simulated and real data.