Select samples randomly based on columns in metadata?

Yes!

First you use
qiime feature-table filter-samples
to make a table for each of your groups of samples, then you randomly split samples into groups using
qiime sample-classifier split-table

While there is a not a tutorial that shows you exacly how to do this, there are docs and discussions about this process.

Let us know if you have any questions!

3 Likes