ANCOM for multiple treatments

Hi everyone, I am very new with Qiime2. Would like to know is that alright for me use ANCOM for comparing multiple soil microbiomes (says microbiomes from 12 different treatments)?

Thank you so much.

Hi @yitkheng,

It is theoretically possible to use ANCOM to test across 12 groups. However, I would make a couple of practical recommendations. First, I would check and see if there’s a difference in beta diversity between groups. If yes, proceed. If no, perhaps reconsider.
Second, I would carefully consider filtering your data. I’d probably look at prevalence here, so what fraction of samples should the feature appear in.

Finally, think about group size. You will struggle to find a signal with 3-5 samples per group.

Best,
Justine

1 Like

Hi @jwdebelius,

Thank you for the prompt response and also the recommendations.
I will try to find out the difference in beta diversity between treatments/groups and also work on the filtering too. Another quick question, for the 12 treatments, we too collected samples from three separate time-points: 0, 2.5, and 5 month-post-inoculation, would it be advisable to possible for us to adopt ANCOM to compare across three-time points for the respective treatment/group?

Thank you.

Cheers

Hi @yitkheng,

This is not supported in the QIIME2 version of ANCOM. It might be supported in the R ANCOM II. You might also check the longitudinal tutorial if you haven’t looked at it and check out the library. I think there’s a shiny new longitudinal plugin there as well. (Although I have not played with said plugin).

Best,
Justine

1 Like

Hi @jwdebelius, I have tried to adopt PERANOVA and ANOSIM using beta diversity, it turned out a few groups were not significantly different. I will try the filtering function and also look into ANCOM2 as well as longitudinal tutorial.

Thank you.
Yit

Hi @yitkheng,

If you don’t have a difference in beta diversity, do you expect to see a difference in ANCOM? Why or why not? Just some food for thought.

Best,
Justine

3 Likes

Hi @jwdebelius, based on the permanova analysis, the treatments without differences were not used for ANCOM. So, far ANCOM does give the taxa that are differentiating between treatments.

Thank you.
Yit

1 Like

Hi @yitkheng,

I’m glad you’re finding signal and differentially abundant taxa!

Best,
Justine

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.