Which reads does DAD2 use to create its' error profiles?

I am wanting to understand the DADA2 process better when used in QIIME2 2019.10.
Does it use the raw imported reads (with primers removed)?
Does it use the raw imported reads after they have been trimmed according to the parameters --p-trim--, --p-trunc-len-*?
Something else?

Hi @minardsmitha,

Yes, it applies the selected trim/trunc settings, performs a pre-filter to remove reads with more errors than the selected number, and then uses the remainder to train its error model and denoises these.

For more details on how the error model is trained and used for denoising, I recommend reading the original publication of this method, Callahan et al. 2016 Nature Methods.

Good luck!

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