Visual Data Mining of Coexpression Data to Set Research Priorities in Cardiac Development Research

Over the past decade, an immense amount of biomedical data have become available in the public domain due to the development of ever-more efficient screening tools such as expression microarrays. To fully leverage this important new resource, it has become imperative to develop new methodologies for mining and visualizing data to make inferences beyond the scope of the original experiments. This need motivated the development of a new freely available web-based application called StarNet (http://vanburenlab.medicine.tamhsc.edu/starnet2.html ). Here we describe the use of StarNet, which functions primarily as a query tool that draws correlation networks centered about a gene of interest. To support inferences and the development of new hypotheses using the resulting correlation network, StarNet queries all genes in the correlation network against a database of known interactions and displays the results in a second graph and provides a statistical test of Gene Ontology term enrichment (keyword enrichment) to provide tentative summary functional annotations for the correlation network. Finally, StarNet provides additional tools for comparing networks drawn from two different selected data sets, thus providing methods for making inferences and developing new hypotheses about differential wiring for different regulatory domains.