Talk Title :
Reconstructing gene networks by epistatic analysis of phenotype and expression data
Date / Time / Location:
Thursday October 8th, 2009 – 6:00 pm
McGill University
Room 232, Leacock Building
855 Sherbrooke Street West
Affiliation :
Ottawa Institute of Systems Biology
URL
Mads Kaern
Abstract :
One the most successful methods to map genetic networks and pathways predates modern genomics by nearly a century. This method, epistatic analysis, traditionally involves observing the phenotypic impact of mutating two different genes individually and in combination. This talk will introduce the basic concepts involved and discuss recent advances emphasizing the inference of transcriptional regulatory networks. In one study, we investigated if conventional epistatic analysis might be used to highlight interactions within the network regulating the transcriptional response of yeast to DNA damage, and developed a method to specifically identify dynamically modulated functional relationships. Somewhat surprisingly, the network derived from phenotypic data has only a modest overlap with that inferred from microarray data. In a second study, we directly compared epistatic analysis based on phenotypic data and marker gene expression. This analysis demonstrates that the two approaches provide complementary information. While conventional analysis correctly infers the order of genes in metabolic pathways, expression-based analysis specifically highlight regulatory hierarchies. Moreover, combining the two methods allows for a nearly complete network reconstruction with a negligible false discovery rate.