Hi, I am interested in showing the effect size for my differential abundance testing. I understand that the coefficient in the Maaslin2 output is an effect estimate, but is it possible to convert it into a more interpretable form such as odds ratio which the clinicians tend to like? Thanks for any advice.
Hi @jjsimz - odds ratio is usually calculated when the response variable (Y) is binary which is clearly not the case with MaAsLin 2. The closest biologically meaningful effect size you can get from MaAsLin 2 is the log2 fold change which can be obtained from the coefficient estimates themselves (see here for an example).
1 Like
Thanks for the response! Log2 fold change works fine.