To choose the best Beta-div comparison approach


I want to compare the microbiota profile of the following stool samples: two groups of mice (KO vs WT), both followed across time (at weeks 3,5,10,20, 56 and 70). How do we perform better the beta div comparison?
We want to see if there are differences at different times between WT and KO (3 vs 3, 5 vs 5, 10 vs 10, and so on). I ckecked the longitudinal tutorial, but I’m not sure about if it fits well to the purpose of the study. I also try to analyze them separately, for example: I generate new variables called Time3, Time5, Time10, etc, with two values: KO or WT, and performed beta-group significance individually on each variable. Is this approach correct? Any advise?

Thanks a lot

Hi @KirKara,

I think you can do a cross-sectional approach, but that you’ll get the most power out of going longitudinal. The microbiome is highly specific to an individual (and less so to a cage), and your longitudinal design lets you harness that power. So, I would focus on the q2-longitudinal pluging and the LME models.

You could also try using qiime diversity filter-distance-matrix to get time-period by time-period differences. It will give you more focused comparisons that a metric in the full table, and let you do per-timepoint PCoAs, if that’s something that you’re interested in.



Thanks a lot for your answer Justine. I’ll try longitudinal analysis.