need to recompute IC after dropping evidence codes

[dkainer]

Hi, thank you for this package. It is very useful!

This is not a bug but a potential issue.

IC is calculated on the full set of Gene-GO annotations at the point when the annotation DB is loaded in the godata() function. If the user decides to drop certain evidence codes (e.g. IEA) during analysis, then the resulting GO annotation set is changed, sometimes quite drastically, and therefore the IC semantic similarity calculations will be using IC scores that are not correct with respect to the final annotations.

Should IC be re-calculated if the user drops codes?

cheers
David

[EEmanetci]

Hello,
I also have same problem, I downloaded the db data seperately for MF,CC and BP and I want to perform similarity analysis with my list of GO terms. but even I used the different db (MF for example), I got the same results. I need to clean the workspace and recalculate the similarity scores seperately. I think there is a problem within the code.
Elif.