z0on
commented
3 months ago Hi Marek - thanks for using my tool! I don’t recommend bringing any measure of significance into this, since it creates bias towards highly expressed genes. Even a tiny fold-change in them is significant because counts are so large. You basically get ribosomal proteins all the time in your GO summaries. Log-fc is a better one: it has its own problem of being noisy for low-expressed genes, but at least the direction of the noise is random so it should average out across all genes in a GO category. Misha
On Tue, May 7, 2024 at 3:25 AM mewojciechowski @.***> wrote:
Hi Mikhail,
Thank you for providing and maintaining the tool. I really appreciate it. I have a question regarding the ranking metric for running GO_MWU. Do you suggest using logFC or a qvalue with a sign added based on logfc as the ranking measure for the genes? Or maybe some more elaborate measure such as -log10(p-value) * log2(foldchange)? What's your current take on that.
Best wishes
Marek
— Reply to this email directly, view it on GitHub https://github.com/z0on/GO_MWU/issues/21, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABZUHGEGNKPTIRFDWYJJR5LZBCFY7AVCNFSM6AAAAABHKP6DYSVHI2DSMVQWIX3LMV43ASLTON2WKOZSGI4DENRRGMZTENY . You are receiving this because you are subscribed to this thread.Message ID: @.***>
-- cheers Misha matzlab.weebly.com
mewojciechowski
Hi Mikhail,
Thank you for providing and maintaining the tool. I really appreciate it. I have a question regarding the ranking metric for running GO_MWU. Do you suggest using logFC or a qvalue with a sign added based on logfc as the ranking measure for the genes? Or maybe some more elaborate measure such as -log10(p-value) * log2(foldchange)? What's your current take on that.
Best wishes
Marek