Interpretation of results of SOM analysis of microarray data by principal components

Motivation: Microarray technology provides a massively parallel means to study gene expression on a global scale. There are many challenges associated with the analysis of microarray data due to its inherent complexity and high dimensionality. Although there is a diverse range of analytical techniques available for finding groups in gene expression data, clustering and partitioning are currently the key areas of microarray data mining. Combining the analytical techniques could provide new ways to improve grouping quality and interpretability. Results: We applied the method of principal components to a united sample of gene expression profiles, presented by Borovecki et al. (2005), and the centers of SOM clusters that we calculated. This allowed us to give a meaningful interpretation to the clusters obtained.