Differential abundance analysis over time with paired samples

Hey guys,

I have the following experimental setup and I am quite unsure what’s the proper tool to use for my problem:

I am looking at the spatio temporal variation of root microbiota and have several field experiments and currently look at my data. Basically, in one of my simpler experiments I have sampled 6 trees 12 times over the course of one year and thus the samples are paired. I have done DEICODE and PERMANOVA for beta diversity but wanted to check which taxa are differentially abundant at different timepoints or whether the individual trees behave differentially. I have already checked out. and tried ANCOM (but there is no option to group my samples), aldex2 in R, but paired test is only available for two timepoints as far as I understood; q2 longitudinal (which does not give. me the possibility to do statistical analysis to my knowledge), ANCOM-BC in R (which I did not fully grasp yet to admit) and looked at songbird (which also has no statistical test included to my knowledge).

Furthermore, most tests look at factors over time but my focus is the time series itself. Any ideas. and help would be greatly appreciated!

Cheers, Max

1 Like

Hi @mfbeuq , I can’t comment on q2-longitudinal (CC @Nicholas_Bokulich ).

You’re right, Songbird doesn’t have support for statistical tests; but again you do need to be cautious about the statistical tests that are returned by most diff abundance tools. Its generally a good idea to save these statistical tests for beta-diversity and use the differential abundance to highlight meaningful taxa.

ANCOM-BC does have support for repeated measures in addition to global test statistics, so I do recommend checking that out.

4 Likes