Power calculation on ANCOM results (ie., are the negative findings related to a beta (type II) error).

Greetings!

In my research article submitted to a journal, the question is asked on ANCOM results like as below.
“Whether a power calculation was performed (ie., are the negative findings related to a beta (type II) error)”

Please, can anyone help me to understand and answer this question regarding ANCOM results?

Hi @srini,

At this point, microbiome power calculations are a loaded question, and ANCOM is a somewhat hairy problem. There is a power calculation addressing a different model for feature-based analysis, but there is not - as far as I know - a power calculation around ANCOM. That would, of course, make it hard for you to perform a power calculation. But, you might to check out Weiss et al and the ANCOM paper for some more of the nitty gritty details around ANCOM.

That said, whether or not you are powered depends a lot on your study design, research question, and desired effect size. Given that GWAS case-control studies - which are orders of magnitude simpler than microbiome - don’t approach fully powered until they hit tens of thousands of people, its probably fair to assume that unless you also have tens of thousands of people, you are underpowered to detect things with very small effect sizes.

Best,
Justine

3 Likes