%0 Journal Article
%T Empirical validation of the S-Score algorithm in the analysis of gene expression data
%A Richard E Kennedy
%A Kellie J Archer
%A Michael F Miles
%J BMC Bioinformatics
%D 2006
%I BioMed Central
%R 10.1186/1471-2105-7-154
%X The S-score showed excellent sensitivity and specificity in detecting low-level gene expression changes. Rank ordering of S-Score values more accurately reflected known fold-change values compared to other algorithms.The S-score method, utilizing probe level data directly, offers significant advantages over comparisons using only probe set expression summaries.Affymetrix GeneChip£¿ microarrays are the most widely used and best standardized platforms for large-scale analysis of gene expression data [1,2]. Current chips are capable of measuring essentially whole genome expression values (>3 ¡Á 104 genes) simultaneously. The Affymetrix technology uses a set of probe pairs, typically 11 to 20 in number, to represent a gene [3,4]. Each probe in the probe pair is 25 bases in length. The perfect match (PM) probe corresponds exactly to the transcript of interest. The corresponding mismatch (MM) probe in the probe pair differs only in the middle (13th) base and is intended to measure nonspecific binding [3,4]. Prior to class comparisons, typically the signal intensities for the probe pairs in a probe set are condensed into an expression summary value, a measure representing the abundance of the corresponding gene transcript [1-3,5]. Statistical tests are then applied to these probe set expression summaries to identify which genes should be declared as differentially expressed [6].Such an approach reflects the two central goals of statistics, estimation and inference. Although usually considered in tandem in microarray data analysis, the two steps are potentially separable [6]. The purpose of most microarray experiments is to draw inferences regarding changes in expression for a large number of genes, and estimating the level of gene expression per se is rarely of interest. The intermediate step of estimating expression summaries may introduce a source of variability to the analytical process, which in turn may affect error estimates used in hypothesis testing. A direct test of
%U http://www.biomedcentral.com/1471-2105/7/154