Comparing Affymetrix Human Gene 1.0 ST preprocessing methods on tissue mixture data

Preprocessing is an essential step of gene expression level estimation in microarray data analysis. Although there are several benchmark studies comparing different preprocessing methods for several Affymetrix platforms, no such studies were performed for the Human Gene 1.0 ST microarray which are substantially different from the previous arrays. We aimed to compare several preprocessing methods on the tissue mixture data provided for the HuGene 1.0 ST platform to assess which methods perform best on this array. The key is the use of gene tissue specificity, tissue proportions and fold change estimates for creating assessment criteria for preprocessing methods. While on some stages of preprocessing it is possible to select uniformly best method, for the others methods could have different ranks according to different criteria.