Gene set testing, stem cells and breast cancer
by Gordon K Smyth
Abstract: In genomic analyses of gene activity, a key problem is to detect genes which change in activity levels between conditions. This leads to a very high dimensional version of model fitting and hypothesis testing. This talk will consider a variant of the problem in which we want to test whether a set of co-regulated genes changes as a group, attaching one p-value to whole set instead of doing genewise tests. This poses challenging statistical problems in that the correlations between the genes are unknown and the sample sizes are tiny. In this talk, we develop three approaches to gene set testing and apply them to solve an iconic problem in cancer biology, identifying the cell of origin for the most invasive form of breast cancer.
For More Information: contact: Mihee Lee. email: email@example.com