How to use qiime group_significance.py script?

Question

[agata88]

Hi all!

I know that maybe this is not a question for this forum but I will be glad if somebody could give me an advice in that.

I would like to compare bacteria abundances (otu_table.biom, qiime) for two groups of samples:

• 35 samples with disease

• 15 controls

I am not sure which test from group_significance.py script should I use. I have a clinical data that need to be correlated with and between those groups.

Do you think that I need to check data for normal distribution? (for every question from clinical data)

Or can I just go with non-parametric test? (eg. Mann-Whitney test).

What test do you use in similar situations?

I will be happy for suggestions.

Best,

Agata

[sridhar56]

I have never tried the group_significance.py but I guess the UNIFRAC comparison will give you a good idea on your comparison,

http://qiime.org/scripts/beta_significance.html

If you have done the pick_de_novo pipeline, you will have the outputs for beta significance.

[agata88]

I don't think this will help ... especially when I need to compare eg 35 vs 15 samples. I think I need to use group_significance.py script with specified mapping file. I just don't know if I need to check data for normal distribution. Can I assume that all data is not normally distributed? (because of small number of sample size) and do the non-parametric test ...

[sridhar56]

Actually I got that wrong, the one I pointed to you was unweighted, you can do the beta diversity which requires a mapping file,

Try this,

http://qiime.org/scripts/beta_diversity.html

I have analyzed many samples this way and through this script, you can have unequal number of samples as long as there are 2 groups.

[agata88]

This looks good ... I think I am going to use that first, thank you :)