Integrating personalized gene expression profiles into predictive disease-associated gene pools

Abstract

Molecular Medicine: Personalized expression profiles elucidate disease heterogeneity A comparison of gene expression profiles of different patients with the same disease reveals a combinatorial model for disease heterogeneity. A framework is developed to generalize group-wise differential expression profiles to personalized perturbation profiles (PEEPs) for individual subjects. It is shown that similarities among patients with the same disease cannot be attributed to a few widely shared genes, but arise from more complex patterns of pairwise overlaps that often correspond to perturbations within the same biological pathway. This points to the existence of a disease module, i.e. a broader group of functionally related genes whose perturbations are associated with the specific disease. The overlap of an individual’s PEEP with the disease module accurately predicts disease status, suggesting their potential for developing combinatorial biomarker signatures or identifying most widely effective drug targets.