Hello everyone,
I performed beta diversity indexes. for example in bray curtis I can see that for example group A is more similar to group B. but how can I have the p-value for these indexes? I mean how can I be sure that the difference is significant or not? or how much are they similar together exactly?
Thank you
Hello!
You can perform PERMANOVA test to check if there are significant differences between your groups of samples. If there are more than 2 groups, you also can perform it in a pairwise mode, so you will be able to find out which groups are different between each other and which are not. In the last case p-values will be automatically adjusted for multiple testing (q-values) 
Here is the link to the plugin:
https://docs.qiime2.org/2021.2/plugins/available/diversity/beta-group-significance/
Best,
2 Likes
thank you very much. yes I did it. but the overall p-value is significant but in pairwise none of them are significant. how do you interpret this?
Also, I used Shannon diversity as well. and the conclusion is that there is a significant difference in microbial community richness between the samples.
on the other hand, if permanova results are not significant, how do we interpret this?
How can two communities have large differences between their richness (Shannon result), but be similar in OTUs (permanova result)?
Probably, all significant p-values disappeared after FDR corrections. If some q-values are lower than 0.1 you still may consider it as a trend toward significance.
Shannon entropy belongs to the alpha diversity group and estimate a diversity of microbial community within a sample, meanwhile Bray-Curtis is a beta diversity metric and estimate dissimilarity between samples. So, you have some differences in the richness within samples between your tested groups, but microbial communities of samples within each group are equally distant from each other as they are distant from samples from another groups of samples
2 Likes
thank you for your great information. so, for example, the q-value is about 0.08, should I consider it significant?
No, it is not a significant result, but you still can discuss or report it as some kind of tendency to significance. Like a trend
1 Like
This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.