1265 Welch Rd
Stanford, CA 94305
USA
Winn Haynes, PhD Candidate,
Khatri & Utz Labs,
Biomedical Informatics Research,
Stanford University,
Abstract:
Disease research has focused on the genes which are most well-characterized instead of the genes with the strongest molecular effects. In our gene expression meta-analysis of over 36,000 patient samples from 103 diseases, we find that only 20% of the top five gene associations for each disease exhibit significant differential gene expression.
Our gene expression meta-analysis improves our molecular understanding of diseases and their relationships. By integrating this gene expression analysis with the electronic health records of two million patients through meta-correlation, we develop a hierarchy of diseases based on patients’ molecular and clinical manifestations. We identify surprising pieces of the meta-correlation hierarchy and pairs of diseases with significant meta-correlation values that were not previously captured in knowledge-based representations of disease. We validate the clinical utility of our analysis since diseases which share drugs are significantly enriched for positive meta-correlation values.