Dada2: Self-consistency loop terminated before convergence

[Stream_biofilm (Danny W)]

Hello Everyone,

I am running a rather large 16S data set using dada2 and the following command:

qiime dada2 denoise-paired --i-demultiplexed-seqs 16S_demux-paired-end.qza --o-table 16S_dada2_table --o-representative-sequences 16S_rep-seqs --p-trunc-len-f 250 --p-trunc-len-r 238 --verbose --p-n-threads 0 --p-n-reads-learn 2000000

I received this warning:

Warning message:
In dada(drpsF, err = NULL, selfConsist = TRUE, multithread = multithread) :
Self-consistency loop terminated before convergence.

I read on Github that this is not always a problem but you should check the error rates and see how close to convergence they are. Is this a function I can do in Qiime2 or only in the dada2 R package? Should I ignore it and continue on? Any suggestions or clarification would be great!

Cheers
Danny

[thermokarst]

Hi @Stream_biofilm! Great question! As you mentioned, there are already some suggestions about the convergence on the official DADA2 docs. Unfortunately, q2-dada2 currently isn’t wired to handle the plotError visualization (also, denoise-single and denoise-paired don’t expose the MAX_CONSIST parameter, so even if you could plot the errors, there is no way through q2-dada2 to make any changes). We have an open issue to track this integration, so stay tuned! In the meantime, you could always poke around at your data directly in DADA2. Thanks!