# How can I do this

**URL:** <https://forum.mothur.org/t/how-can-i-do-this/1943>\
**Category:** Commands in mothur\
**Created:** [July 26, 2014, 5:28pm UTC](https://forum.mothur.org/t/how-can-i-do-this/1943 "2014-07-26T17:28:40Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![sdpapet](https://avatars.discourse-cdn.com/v4/letter/s/898d66/32.png) [@sdpapet](https://forum.mothur.org/u/sdpapet)\
**Post date:** [July 26, 2014, 5:28pm UTC](https://forum.mothur.org/t/how-can-i-do-this/1943/1 "2014-07-26T17:28:40Z")

</div>

Hi, I want to compare the community percentage composition between two groups.  
For example, the data set like this.  
site 1 site 2

Acidobacteria 30% 25%

Actinobacteria 24% 20%

Cyanobacteria 5% 6%  
.  
.  
So on and so forth.

Site 1 and Site 2 can be two sites or the same site sampled in different seasons (e.g. spring, summer). I want to get some information if the community composition changed or not.

I was wondering which analysis/statistical methods can compare them directly.

The methods that i can think about is ANOVA?. Normally, if I use ANOVA, each site has to have multiple measurements (e.g., I have more than 2 sets of composition data in site 1 and site2)  
Any command in mothur can do this? If not, can any one tell me which stats method can do this, I can do it in R.

Can fisher exact test or G-test do this? (I don’t think these two tests can use percentage data)

Thanks,

---

<div class="post-metadata">

**Author:** ![dwaite](https://avatars.discourse-cdn.com/v4/letter/d/ecccb3/32.png) [@dwaite](https://forum.mothur.org/u/dwaite)\
**Post date:** [July 26, 2014, 9:56pm UTC](https://forum.mothur.org/t/how-can-i-do-this/1943/2 "2014-07-26T21:56:06Z")

</div>

You can to this with metastats ([command](http://www.mothur.org/wiki/Metastats) and [paper](http://www.ploscompbiol.org/article/info%3Adoi%2F10.1371%2Fjournal.pcbi.1000352)).

You’ll need to create an appropriate shared file using phylotyping to get the phylum-level groupings, and then manually create your design file to attribute samples to site groupings, but the phylotyping is quite simple and design file just takes a few minutes in any text editor.

Hope that helps.

---

<div class="post-metadata">

**Author:** ![sdpapet](https://avatars.discourse-cdn.com/v4/letter/s/898d66/32.png) [@sdpapet](https://forum.mothur.org/u/sdpapet)\
**Post date:** [July 26, 2014, 10:34pm UTC](https://forum.mothur.org/t/how-can-i-do-this/1943/3 "2014-07-26T22:34:51Z")

</div>

Thanks. Can you tell me attach or tell me where can I find your example design file

design=temp.job2.design

---

<div class="post-metadata">

**Author:** ![dwaite](https://avatars.discourse-cdn.com/v4/letter/d/ecccb3/32.png) [@dwaite](https://forum.mothur.org/u/dwaite)\
**Post date:** [July 27, 2014, 3:40am UTC](https://forum.mothur.org/t/how-can-i-do-this/1943/4 "2014-07-27T03:40:42Z")

</div>

[http://www.mothur.org/wiki/Design\_File](http://www.mothur.org/wiki/Design_File)

It’s just a two-column, tab-separated file with group names on the left, then treatments or metadata factors on the right. For example:

Group1 Summer  
Group2 Summer  
Group3 Summer  
Group4 Winter  
Group5 Winter  
Group6 Winter