# Longitudinal analysis of beta diversity
**URL:** https://forum.qiime2.org/t/longitudinal-analysis-of-beta-diversity/11326
**Category:** User Support
**Created:** [August 30, 2019, 1:21pm UTC](https://forum.qiime2.org/t/longitudinal-analysis-of-beta-diversity/11326 "2019-08-30T13:21:33Z")
**Posts on this page:** 4
**Page:** 1
### Author:  [@m\_max](https://forum.qiime2.org/u/m_max)
#### Post date: [August 30, 2019, 1:21pm UTC](https://forum.qiime2.org/t/longitudinal-analysis-of-beta-diversity/11326/1 "2019-08-30T13:21:33Z")
Hi
I have a little question for the community. I’m performing a study, involving microbiota composition, on a population of samples in which I have diseased subjects and “healthy” controls. The diseased subjects were sampled at two different times (T0 and T1). I calculated alpha and beta diversity (e.g Shannon index and Bray-Curtis) for diseased group and performed comparison between the alpha diversity of samples at T0 and T1 by using “qiime longitudinal pairwise-differences”. I would like to perform the same analysis for beta-diversity determining if a modification in microbiota composition between the two time-points happened. I tried to use “qiime longitudinal pairwise-distances” but it requires to specify at least two group while I have only one group (diseased) sampled at two time-points. Any suggestions???
Thanks for your help!!
---
### Author:  [@Nicholas\_Bokulich](https://forum.qiime2.org/u/Nicholas_Bokulich)
#### Post date: [August 30, 2019, 1:27pm UTC](https://forum.qiime2.org/t/longitudinal-analysis-of-beta-diversity/11326/2 "2019-08-30T13:27:38Z")
Are the healthy controls only sampled at one time point?
> [@m\_max](#):
>
> I tried to use “qiime longitudinal pairwise-distances” but it requires to specify at least two group while I have only one group (diseased) sampled at two time-points.
We have an [open issue](https://github.com/qiime2/q2-longitudinal/issues/114) to make the "group column" optional in that action. The reason it has not been implemented sooner is that pairwise-distances on a single group is not too informative — the distances are always positive and almost always \> 0, so this test is really only useful for demonstrating that one group changed _more_ than another group did over the same time periods. Contrast that with pairwise-differences, for which it is actually meaningful to show that, e.g., alpha diversity significantly increased or decreased for a single group.
There is another possibility for you: calculate the PCoA coordinates and use those as input for pairwise-differences. Would that accomplish what you are after?
---
### Author:  [@m\_max](https://forum.qiime2.org/u/m_max)
#### Post date: [August 30, 2019, 2:58pm UTC](https://forum.qiime2.org/t/longitudinal-analysis-of-beta-diversity/11326/3 "2019-08-30T14:58:14Z")
Hi Nicolas
I agree with you that the distances in pairs on a single group are not too instructive. Unfortunately those who performed the experimental design did not think to sample the checks at the second time (I think it is indispensable). I will try to follow your suggestion with the PCoA coordinates.
Thanks for your help
---
### Author:  [@system](https://forum.qiime2.org/u/system)
#### Post date: [September 30, 2019, 8:58pm UTC](https://forum.qiime2.org/t/longitudinal-analysis-of-beta-diversity/11326/4 "2019-09-30T20:58:16Z")
This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.