I’m trying to figure out an elegent way to apply double demultiplexing on samples which have already been demultiplexed once. (Similar to the Ion Torrent/Multiple regions) problem. (Maybe @cjone228 and @Lauren have already solved this, and I haven’t found the thread.)
The inelegent solution Ive come up with is to import each sample as a multiplexed sample, demultiplex by region, filter each sample into regions, and then try to find a way to merge the sequences before running dada2.
qiime tools import \
--input-path example-seqs/ \
--output-path example-all.qza \
--type 'MultiplexedPairedEndBarcodeInSequence'
qiime cutadapt demux-paired \
--i-seqs ./example-per-sample-seqs.qza \
--m-forward-barcodes-file barcodes.tsv
--m-reverse-barcodes-file barcodes.tsv \
--m-forward-barcodes-column fwd-barcode \
--m-reverse-barcodes-column rev-barcode \
--o-per-sample-sequences example-by-region.qza \
--o-untrimmed-sequences example-unmapped.qza
At which point, i can seperate things back out into per-region files and then loop back into QIIME 2 by exporting the artifact, building a per-region manifest, and then re-importing. (Given my occasionally questionable typing skills, there is a non-zero probability this will lead to an error between and ).
So, one thought for larger studies would be to have a cutadapt method that worked on a per-sample basis and then seperated the samples back out to a set of SampleData[SequencesWithQuality]
for denoising, etc.
Again, Im not sure if the problem has been solved, but Im struggling with a solution that doesn’t require several data-related liberties.
Best,
Justine