New software moves microarray analysis from individual genes to whole networks of genetic interactions

A graphical interpretation of the GSEA methodImage courtesy of Broad Computational Biology and Informatics

A group of scientists led by Broad researchers have given the biomedical community a powerful new analytical tool to help understand the subtle genetic alterations that underlie most complex diseases. Described in the October 25 issue of the Proceedings of the National Academy of Sciences, Gene Set Enrichment Analysis (GSEA) moves microarray analysis from the level of individual genes to whole networks of genetic interactions.

GSEA can be applied to diseases, such as diabetes and cancer, that are difficult to understand because the genetic alterations are subtly different across the many different interacting genes. It is only when these alterations are viewed as a group — and measured as a group with the new statistical technique embedded in GSEA — that a clear signal can be detected above the background "noise."

"The power of analyzing molecular profiles with GSEA is that it allows us to identify subtle signals, derive more invariant signatures, and, by using gene sets associated with biological processes, to begin to get more understanding of the underlying mechanisms of both diseased and healthy cellular states," says Jill Mesirov, senior author and director of Broad Computational Biology and Bioinformatics.

Although the tool itself is powerful, the backbone of this approach is a well-curated database of gene sets, i.e., genes that are known to interact in biochemical pathways or "molecular signatures." To date, the researchers have gathered ~1500 such sets, and made them publicly available with the GSEA software package. The fact that this database will continue to grow as further molecular interactions are defined in the scientific literature gives GSEA an unparalleled potential in finally understanding the collective behavior of genes in states of health and disease.

An elegant proof of the power of this approach is also described in the PNAS paper. The researchers tested their tool on several cancer-related data sets. In one particularly compelling analysis, they reanalyzed data from three recent lung cancer studies from three different labs. The previous studies looked at gene expression profiles in lung adenocarcinoma patients with associated clinical outcome endpoints. The Broad group's analysis found no significant overlap in single gene alterations between the studies. However, when whole gene sets were analyzed in the three studies, a strong common signal that correlated with outcome was uncovered. Such analyses not only point toward diagnostic advances, but also suggest new areas of focus for understanding and attacking disease.

"Gaining insight into the biological or mechanistic messages contained within expression-based array experiments has been notoriously difficult," says Broad associate member William Sellers. According to Sellers, GSEA has proved quite successful in this area and was instrumental in delineating a role for the protein mTOR in regulating hypoxia-inducible factor-1 (HIF-1) activity, as well as the activity of glycolytic pathways downstream of HIF. HIF regulates gene expression in solid tumors and influences both angiogenesis and tumor growth. "These findings have allowed us to look in patients using PET scans to measure pathway activity and try to determine whether the drug is having an effect in the patient " says Sellers, who is an associate professor of medicine at Harvard Medical School and a clinician at the Dana-Farber Cancer Institute.

In keeping with Broad's mission of creating genomic tools and making them available to the research community, GSEA may be freely downloaded at www.broadinstitute.org/gsea.

About the Broad Institute of MIT and Harvard

The Broad Institute of MIT and Harvard was founded in 2003 to bring the power of genomics to biomedicine. It pursues this mission by empowering creative young scientists to construct new and robust tools for genomic medicine, to make them freely accessible to the global scientific community, and to apply them to the understanding and treatment of disease.

For further information about the Broad Institute, go to http://www.broadinstitute.org.