Good morning,
I have seen on different data set that the LefSe running inside a conda environment returns less results than LefSe running on Galaxy (computed on the same data set with the same settings, normalization included). The cause seems to be the LDAscore step and not the pre-selection step (because here there are the same results on both Conda and Galaxy, and the wilcoxon does NOT filter any feature as set).
Getting less results clearly it’s not an issue but I would like to appreciate what’s the cause of this difference.
Thank you,
Leandro