The identification from global data sets of stable and predictive patterns of relative expression reversals offers a simple, yet powerful approach to target therapies for personalized medicine and to identify pathways that are disease-perturbed. We have utilized this approach to identify a molecular classifier (Price et al, PNAS, 2007) with near 100% accuracy for differentiating gastrointestinal stromal tumor (GIST) and leiomyosarcoma (LMS), two cancers that have very similar histopathology, but require very different treatments. We have also developed a novel method, called pathway eXpression Rank AnalYsis (p-XRAY) for utilizing multiple relative expression reversals between genes within a priori defined gene sets that are informative about pathways that are perturbed and differentially regulated between phenotypes. The method is noteworthy because it 1) is independent of data normalization; 2) results in an elegant classifier where binary phenotype (e.g. disease) diagnosis can be done based simply on whether the metric is computed to be above or below zero (and is bounded by 1 and -1); and 3) appears thus far to be more accurate than current state-of-the-art pathway comparison methods. We have also used the relative expression reversal strategy to develop a suite of molecular classifiers that can diagnose a large number of brain diseases (many types and stages of brain cancers, Alzheimer's, Parkinson's, Epilepsy etc.) from transcriptomic measurements. Such molecular diagnostics to successfully distinguish between not just a disease and a control group, but among many different phenotypes simultaneously are critical to enable systems approaches for personalized medicine.

Extended Abstract Status: Not Uploaded