Correlation-based Data Representation

The Dagstuhl Seminar Similarity-based Clustering and its Application to Medicine and Biology (07131) held in March 25–30, 2007, provided an excellent atmosphere for in-depth discussions about the research frontier of computational methods for relevant applications of biomedical clustering and beyond. We address some highlighted issues about correlation-based data analysis in this seminar postribution. First, some prominent correlation measures are briefly revisited. Then, a focus is put on Pearson correlation, because of its widespread use in biomedical sciences and because of its analytic accessibility. A connection to Euclidean distance of z-score transformed data outlined. Cost function optimization of correlation-based data representation is discussed for which, finally, applications to visualization and clustering of gene expression data are given.