qiime cutadapt demux failed to exclude reads with fake barcodes in the middle

[Ant_Wisdom]

Below is my command.
qiime cutadapt demux-paired --i-seqs asap2_out/imported/fqMuBiPe-1.qza --m-forward-barcodes-file input//fqMuBiPe-1/metadata.tsv --m-forward-barcodes-column barcode-sequence --o-per-sample-sequences asap2_out/demultiplexed/fqMuBiPe-1-demux.qza --o-untrimmed-sequences asap2_out/demultiplexed/fqMuBiPe-1-untrimmed.qza

The output has many partial reads because there were reads (e.g. reverse complementary, or contaminants) with no barcode at the beginning, but with a fake barcode (randomly matched) in the middle, and qiime mistook it as a barcode and trimmed it and assign it to a sample.

I temporarily use the parameter --p-minimum-length 225 to compromise it, but a better way should be making qiime to limit the location of barcodes to the first n (e.g. 10) bases. Any ideas?

[colinbrislawn (Colin J Brislawn)]

@Ant_Wisdom

a better way should be making qiime to limit the location of barcodes to the first n (e.g. 10) bases. Any ideas?

There's a lot of cutadapt settings, so there my be multiple ways to do this.

  --p-times INTEGER       Remove multiple occurrences of an adapter if it is
    Range(1, None)        repeated, up to `times` times.          [default: 1]

That's 1 by default, so I'm not sure why it would grab a second barcode...

Also consider --p-front-f using ^sequence

Sequence of an adapter ligated to the 5' end. The
adapter and any preceding bases are trimmed. Partial
matches at the 5' end are allowed. If a ^
character is prepended, the adapter is only found if
it is at the beginning of the read. Search in
forward read. [optional]

This also caught my attention:

The output has many partial reads because there were reads (e.g. reverse complementary, or contaminants) with no barcode at the beginning

I've been given runs like that. Any chance of reamplifying these samples?