Existing and emerging genome-wide association (GWA) datasets, merged with medical record or survey data, enable testing for associations for dozens of phenotypes, yet methods for characterizing the shared genetic architecture of multiple traits are still not well-established. We work on new methods which are based on hierarchical clustering of gene-level association test results for characterizing and clustering the genetic architecture of multiple phenotypes. Our goal is to identify groups of phenotypes or "clusters" that share a core set of significant genes which can be detected even in the presence of noise.
Sohini Ramachandran, Director of the Center for Computational Molecular Biology
Bjorn Sandstede, Director of the Data Science Initiative