Hi, I have a dataset here with two groups (WT and HD). I get significant bray-curtis and weighted-unifrac distances but not unweighted-unifrac distances or jaccard. But when I do the ancom test on the dataset, there are no significant differential abundance identified (even at the phyla level) although the two groups show really clear separation on the PCoA plot, so I am quite confused about this: is it possible to have significant beta-diversity distances but no differential abundance between the OTUs? or is ancom too conservative? Thanks in advance. Cheers.
Interesting – not sure what is going on here. Note that one does not imply the other – getting significant differences in beta diversity does not guarantee you’ll get hits in differential abundances (since they are very different methods).
How significant are we talking about? What do the p-values, and test statistics for the beta diversity methods look like?
Here are the significance of the beta-diversity methods:
For bray-curtis methods: p-value is 0.001 and test statistic is 5.60213.
For weighted-unifrac method: p-value is 0.005, test statistic is 4.68767.
For unweighted-unifrac: p-value is 0.008, test statistic is 2.01683
Btw, what does the test statistic indicate?
FYI, I also ran a gneiss analysis and there were no coefficients with p-value < 0.05. So I am not sure what is going on with my data.
Ok - looking over your results, you are an incredibly small sample size. I wouldn’t expect to get meaningful results in differential abundance analysis unless you have a crazy large effect size. From your emperor plots and permanova results, it doesn’t look like you have a crazy large effect size, so it isn’t surprising that you aren’t getting any hits for the differential abundance.
They are likely explanatory taxa explaining your separation - but it is hard to say. More samples would help reach a more solid conclusion.