Hi,
I’m trying to investigate the genetic dependencies of cancer cell lines which harbor an ARID1A mutation vs cancer cell lines that do not have an ARID1A mutation using the RNAi DepMap dataset. On the online web portal, I’ve seen that the DepMap Tool uses an ‘Effect size’ metric on the x-axis to assess genetic dependency in DepMap datasets. I was wondering if anyone knows what method is used to calculate this effect size on the DepMap data explorer (Cohen’s d, Pearson correlation coefficient, something else?)? Would really appreciate any insight on this! I’ve scoured the literature and am left confused.
On the portal “effect size” refers to how the term is used in statistics: Effect size - Wikipedia
As a result, the effect size depends on the specific statistical test used to do the comparisons. If you are looking at a comparison of two continuous variables data explorer, then we are using pearson’s correlation coefficient, and that’s the effect size. If you are you doing a two-class comparison then I believe the effect size is the difference in means between the two groups.
(In the case of the two-class comparison there may also be some adjustment to the means, but I can’t recall off the top of my head. Regardless the statistical test is a moderated t-test, so the difference between means is what gets reported as ‘effect size’ in this context.)
Thanks,
Phil
