Sampling depth and taxonomic analysis


Currently the QIIME 2 tutorial uses the dada2 output table for taxonomic analysis. How do you incorporate a sampling depth into your taxonomy generator to ensure standardization with the alpha diversity output.
Thank you

Hi @Asha,

There is no need to do this, nor is it recommended.

The alpha diversity results are rarefied — normalization is necessary to account for uneven sampling prior to performing any type of statistical comparisons between samples.

Taxonomy results do not require any normalization, largely because these results are used and analyzed in very different ways. With taxonomy data we are not performing the same types of count-based tests that we use for alpha diversity. Instead, compositional analyses (e.g., ancom, gneiss) are used, which perform their own internal normalization.

If you really must rarefy your data, e.g., because you want to see how rarefaction impacts the taxonomic composition of your samples, you can use feature-table rarefy to rarefy at the same level that you are using for alpha diversity. But DO NOT use that feature table as input for gneiss, ancom or other downstream tests.

I hope that helps!


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