QIIME 2 statistical tests

User Support
1

HELLO,

Can you share the script to perform statistical tests (eg U Mann Whitney test) and to calculate p values using QIIME2 ?

Thanks

2

Hi @M_F,

Please see this post. There are several other similar threads in the forum you can search.

1 Like
3

Hi @SoilRotifer

I see the post . It is indicated that " Mann Whitney is not appropriate for differential abundance, and QIIME doesn't support a per-taxon Mann Whitney or Kruskal Wallis test."

Any advice related to statistical tests for microbiome analysis.

4

Hi @M_F,

What do you mean by "microbiome analysis"? There are many ways to analyze microbial data. As you highlighted the part about differential abundance, I assume that is your primary question.

I'd suggest you look into the q2-composition plugin. Check out the ANCOM-BC2 portion of the Gut-to-soil tutorial.

1 Like
5

Hi @SoilRotifer

I used the undermentioned command line

qiime composition ancombc \
  --i-table table.qza \
  --m-metadata-file metadata.tsv \
  --p-formula ..... \
  --p-reference-levels  .....\
  --o-differentials dataloaf.qza
How can i explore the dataloaf.qza file in term of statistical analysis noting that when using QIIME VIEW tool i see bibliographic references displayed !

Thanks
6

Hi @M_F ,

I suggest you run through the rest of the commands in the ancombc2 tutorial I've linked first. Also checkout the other tutorials, as well as search the forum on this topic.

In brief, if using ancombc run qiime composition da-barplot ... If running ancombc2, run qiime composition ancombc2-visualizer ... These will generate interactive visualizations were you can explore your data.

Otherwise, you can export your data using qiime tools export ... to view with other tools etc...

7

Hi @SoilRotifer
Feature identifiers are indicated as (y-axis labels) in barplots when running
qiime composition da-barplot.
Any advice to display directly the bacteria name (y-axis labels) and how to get p values ?
Thanks

8

The only way to see taxa names using ancombc is to run the analyses a little differently. That is collapse your features by taxonomy (qiime taxa collapse ...) and then run ancombc. Though this means many features will be condensed into one taxonomic label... which can provide different results, and may require a different level of interpretation compared to feature-level analysis.

However, the newer ancombc2 allows you to annotate your features with taxonomy that that is your goal.

If you are looking for a table with p and q values, then simply run qiime composition tabulate ... Again, read the help text for the various options by running qiime composition --help.

-Mike

9

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.