Demux results in missing samples from 3 separate libraries


I have a quick question. I have 3 separate libraries that are not demultiplexed and are single-emp. I have imported the files and ran demux (see codes below), but when I check the output of the demux files I notice that each library is missing 3-7 samples. I found a different question here about this and I have confirmed the barcodes and samples names and they all match so I really have no idea what is going on. Can you please let me know how to proceed?


qiime tools import \
--type EMPSingleEndSequences \
--input-path Plate1/RawSeq \
--output-path Emp-SingleEnd-SeqRaw-1.qza


mkdir Plate1/Demux

qiime demux emp-single \
--i-seqs Plate1/Emp-SingleEnd-SeqRaw-1.qza \
--p-no-golay-error-correction \
--p-no-rev-comp-barcodes \
--p-no-rev-comp-mapping-barcodes \
--m-barcodes-file Metadata/MetadataPlate1.tsv \
--m-barcodes-column Barcode \
--o-per-sample-sequences Plate1/Demux/demux-plate-1.qza \
--output-dir Plate1/Demux/demux-plate-1-details.qza

qiime demux summarize \
--i-data Plate1/Demux/demux-plate-1.qza \
--o-visualization Plate1/Demux/demux-plate-1.qzv

Okay, so I reran with out the golay error (--p-no-golay-error-correction \ ) and the 24 samples for one of the libraries were all present. So given the type of barcodes I have, I definitely should not include the golay error correction, correct?

My question now is, if the golay error decided to discard 3 samples, does this mean that this samples are so bad that they should indeed be discarded? Or what exactly does it mean and why does demux keep the samples when I do not include the golay error correction vs when I do not?

Sorry, I know its a lot to ask but I am honestly very confused

Sample ID.                Barcode
|2020-11-18-14RRX-R1-L | GTGGGACTGCGC|
|2021-02-24-18RXX-R1-L | GCATTCGGCGTT|
|2021-02-24-20LRX-R1-L | ACTGGCAAACCT|
1 Like

Hi @fabipc,
I think there is a chance that you are getting those few sequences that are randomly getting assigned to samples but we haven't found the correct orientation of your barcodes to correctly assign sequences to samples.

I see you used these parameters. Have you tried different flags for these params? What were your results if you flip each of those seperately?