Including linear models for ANCOM as it's in gneiss

Dear QIIME2 developers,

I worked through the gneiss tutorial and I found it a great feature for QIIME2 users which allows us to explore different factors that may contribute to the differential abundance at the same time. In supplemental material of the ANCOM paper, authors mentioned that ANCOM can also be used to build linear models. Can linear models be built for the ANCOM as well so that we can explore different exploratory variables at one time?


Hey @yanxianl,

That sounds like a really useful idea!

To do that we would ultimately need to support this in scikit-bio (because the current iteration of this plugin just calls into that library).

@mortonjt would probably be the person to talk to for that. But it’s also my understanding that we’re looking for a future maintainer for q2-composition so we don’t really have a good timeline or really the resources to dedicate to this at the moment.