Dear MA forum
We’ve been using your pipeline and we wondered how you are doing the CLR transformation in the normalization step, as we cannot find this function “clr_transform” in either of the edgeR and metagenomeSeq packages that are loaded and is used in the MA code (xia-lab/MicrobiomeAnalystR source: R/general_proc.R line 378).
This function does seem to exist in seqtools (seqtools/clr.R at master · skembel/seqtools · GitHub) . If this is the function we are concerned that it ignores the zeros in a user’s data when calculating the geometric mean during the CLR transformation – we believe that ignoring the zeros is called a robust-clr – but can you clarify how the CLR transform works and what it does with the zeros?