Better Affymetrix Estimates

Mark Reimers1, Karolinska

Several biostatisticians have recently made great strides in developing robust estimates of abundance based on multiple probes; such methods are used to analyze microarray and proteomics data. We have improved these estimates further by using a spatially heterogeneous normalization procedure, and by assessing probe quality, prior to using a robust method for fit. We illustrate these methods with the Affymetrix spike-in data set, and with several clinical data sets.