Entering edit mode

2.3 years ago

pixie@bioinfo
★
1.5k

Hello, I have some co-expression clusters identified. I wanted to calculate the statistical significance of these correlated genes in a cluster as opposed to a random cluster of genes. To that end, I have calculated the mean pearson correlation for a given co-expression cluster (every gene-to every other gene) so r_cluster1 . Then I have taken a random cluster and calculated the mean r_cluster_random. Then I have done this 1000 times randomization and got r_cluster_random_1k. Obviously, the r_cluster1 is much higher than r_cluster_random_1k. How do I represent this as a p-value or a Z value ?

Is there a specific reason why you opted against using the established Gene Set Enrichment?

I have two things to show to the reviewers: Statistical significance of the network cluster and biological relevence. So for the biological relevence, we are showing GO enrichment. I also have to show that my clustering method is robust. For that I need to compare against a random network