Fritz Roth
Harvard Medical School
I will describe analysis of two systematically collected genetic interaction networks: 1) ~700,000 double-mutant yeast strains and 2) quantitative growth phenotyping of all double-mutants among 26 DNA repair genes. Synthetic sick or lethal (SSL) interactions explain robustness of an organism to mutation. In the first study, we examined a large SSL network for value in predicting gene function, relationships to other biological networks, and the role of transcriptional compensation in robustness to gene loss. In the second study, we evaluated quantitative measures of epistasis for their ability to predict functional relationships. We classified genetic interactions into multiple sub-types, including co-equality, which corresponded closely to genes functioning as cohesive units (e.g. encoding subunits of the same protein complex). We also applied algorithms that characterize order of action in DNA repair pathways based solely on quantitative growth rates.