Upon manual inspection of these clusters, we assigned biological themes to 73 clusters where the theme was clearly identifiable. To do this, we relied on the MSigDB annotations of both the gene sets, in the form of their names and descriptions, and their individual genes. In a number of cases, we also used complementary annotation tools such as DAVID (Huang et al., 2009b) to obtain additional clues about the most relevant pathways represented by them.
