I've brought up this issue before when I was analyzing Phyla data and was told that I should instead look at the genera level. However, I am now receiving W scores of 0 at the genera level. I've attached my .qzv file that includes the provenance to this post. Note that I've made sure to filter to a min frequency of 10 and a min sample of 2 prior to running ANCOM.
I’m not a statistician, but I wanted to ask a little more about the filtering you preformed on this genera level table because I’m a little surprised that this table only includes 16 features. Usually I find 100s or 1000s of features at the genera level, so I’m wondering if some filtering along the way reduced the number of features and may have confused the statistical model.