Hello @Che,
Would there be a problem with just merging the single read demux results from the same sample together?
Not sure what you mean by this. I think you might be confused by the output you're getting. demux-paired outputs demuxed paired-end reads, not demuxed single-end reads. So no merging necessary.
This issue now is the other 10% of reads that you collected with --o-untrimmed-sequences. Not sure what's going on with these. Try grepping for the reverse complement of the barcodes in these. Maybe also look for the primers. Essentially just try to identify what's unqiue about these and why they're not demuxing.