Taxa stability, two time-points

We have examined the microbiota in the airways at two time points. A referee wants us to identify which of the ASVs that are stable/don’t change. His/her suggestion was to “compare the microbiome of those subjects that have little variability vs. the microbiome of those that have high variability (for example defined as below vs. above the median value of beta diversity distance of the within individual pairs). The authors could use LEfSe, DESeq or other form of comparison to explore this.” However, these methods (as I understand it) don’t take the compositionality of the data into account, and are also not very good at handling paired data.

Does anyone have a suggestion on how to look for stable ASVs? And how to test for it statistically?

Grateful for any suggestions!



I’ve recently used ALDEx2 for differential abundance tests, which takes into account the compositional nature of the data. See this for e.g.

A little bit late to the party on this but jut wanted to point out that there is a compositonally aware differential abundance tool that actually tests differential variance in addition to abundance. This might help you identify taxa with high variance which you could interpret as less “stable”. Check out the Corncob R package. There is also a qiime2 q2-corncob plugin but I personally haven’t used that version so I’m not sure how it would compare against the complete R package.

1 Like