MAY 13, 2015 03:00 PM PDT

Using the network architecture of eQTLs to understand complex traits

  • Postdoctoral Research Fellow, Dana Farber Cancer Institute/Harvard University
      John Platig received his PhD in Physics from the University of Maryland. His thesis focused on the applications of complex network methods to biological data sets, with an emphasis on understanding how errors in edge identification affect network properties. In conjunction with his physics training, he was a Cancer Research Training Fellow at the National Cancer Institute, working to identify potential therapeutic targets from a reconstructed gene regulatory network in Diffuse Large B Cell Lymphoma. In 2013 John started as postdoctoral fellow with John Quackenbush at the Dana-Farber Cancer Institute. He is currently working on network inference and clustering methods to better understand genetic and other factors that drive phenotypes.


    Genome Wide Association Studies (GWAS) and eQTL analyses are producing huge numbers of associations and show no signs of slowing. There are now more than 8,500 SNPs associated with more than 350 complex traits reported in the NHGRI GWAS Catalog. However, interpreting these associations collectively in a functional context remains a challenge. Using genotyping and gene expression data from 163 lung tissue samples in a lower respiratory disease study, we calculated eQTL associations between SNPs and genes and cast significant associations as links in a bipartite network. We identified biological function by focusing on densely linked communities, which comprise groups of SNPs associated with groups of genes. By investigating the intermediate scale of network organization, we found GWAS SNPs enriched at the cores of these communities, including GWAS hits for COPD, asthma, and pulmonary function, among others. We believe these methods are widely applicable to any data set that can be represented as a bipartite network with a giant connected component.

    Show Resources
    Loading Comments...