Rarefaction pior to diversity analyzes

[andressa_mv (Andressa)]

Hi! I am beginning to use Qiime2 and I analyzed a HiSeq dataset based on the tutorials. However, I still have some questions about the rarefaction of the data. Since I’d like to have the taxonomic analysis rarefied by the number of sequences, I applied the following command before the diversity analyzes:

qiime feature-table rarefy --i-table table.qza --p-sampling-depth x --o-rarefied-table rarefied-table.qza

However, I have to add the same value again when doing diversity analyzes, since the parameter --p-sampling-depth is required.

I have two doubts:
1 - Is this correct? Is this the best way to rarefied my data before the downstream analysis?
2 - Why does the taxonomic analysis of Qiime2 does not consider this value? Should data from Dada and Deblur be treated differently than 97% OTUs?

Thank you!

[thermokarst]

Hi @andressa_mv, thanks for writing!! I might be misunderstanding your question, but we generally don't use rarefied feature tables for taxonomic analyses, because we typically perform these analyses using relative frequencies (or perhaps some other scaling method), not absolute frequencies for feature counts. For the methods available in q2-taxa, q2-composition, and q2-gneiss, they all generally accept some form of non-rarefied tables (rarefying would/could alter those relative frequencies quite a bit!).

@andressa_mv

Should data from Dada and Deblur be treated differently than 97% OTUs?

Perhaps --- DADA2 produces "Sequence Variants," which are more similar to 100% OTUs. I suggest reading about it!

Thanks!