A Proposal for Gene Signature Integration

Gene expression patterns that can distinguish to a clinically significant degree disease subclasses not only play a prominent role in diagnosis but also lead to therapeutic strategies that tailor treatment to the particular biology of each disease. Nevertheless, gene expression signatures derived through statistical feature identification procedures on population datasets have received rightful criticism, since they share only few genes in common for a particular pathology, even if they derived from the same dataset using different methodologies. An optimistic view to this problem emerging from the wealth of biological interactions is that a statistical solution may not be unique. The derived signatures may be complementary parts of a global one, with each individual signature intersecting only a small part of biological evidence. In this work we focus on the biological knowledge hidden behind different gene signatures and propose a methodology for integrating such knowledge towards retrieving a unified signature.