How do you choose multiple columns when analyzing Beta diversit?


Sorry for the basic question as I'm new in qiime2 and still learning. How can I choose two different columns when measuring beta diversity? I want to compare both the priming and Strain columns in my metadata file. Here is the code I used when selecting only one column:

qiime diversity beta-group-significance
--i-distance-matrix diversity-metrics-results/weighted_unifrac_distance_matrix.qza
--m-metadata-file metadata.tsv
--m-metadata-column Strain
--o-visualization diversity-metrics-results/weighted-unifrac-strain-significance.qzv

I also attached my metadata file

I appreciate your help!

Thank you,
metadata.tsv (626 Bytes)

Hello @Abdul, unfortunately, you can't run qiime diversity beta-group-significance with multiple columns. What you could do that achieves something similar with multiple columns is run qiime diversity adonis with strain + priming as your formula.


Sorry for the late reply! I have run the adonis using the following code:

qiime diversity adonis
--i-distance-matrix diversity-metrics-results/weighted_unifrac_distance_matrix.qza
--m-metadata-file metadata.tsv
--p-formula "Strain * priming"

--o-visualization adonis-WTBcl6-PS.qza

However, I could not interpret the results. I was expecting a pairwise analysis with different priming conditions vs the strain types. I selected the two columns (priming & strain) where I have different priming conditions and strains. I attached the output. is this the correct output of adonis? thanks for help!

1 Like

Yep, that's the ADONIS output.

You can read more about the Permutational Multivariate Analysis of Variance Using Distance Matrices.

R2 is your 'R squared' value and is just like an R2 value you would get from a linear regression, in that it explains what fraction of variance is captured by that predictor variable. (think effect size)

Pr(>F) is your p-value. As is tradition, you get to pick an alpha threshold, with both a=0.05 for 95% and a=0.01 for 99% being common.

:warning: ADONIS partitions your distance matrix and performs tests in order :scream_cat:
I've written about that here: interpretation beta diversity / permdisp / adonis - #4 by colinbrislawn


Thank you so much for the detailed explanation!

1 Like

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