Errors when demultiplexing paired end sequences

Technical Support
1

Hi, I have questions about demux paired-end sequences (qiime2 version 2020.8).
From my three data files -- R1, R2, I1 -- I'm now trying to demux the sequences.
Although I have manually checked and compared that the sequences in my barcode file and mapping file are in the same orientation, using only the following parameters will result in an error message saying "no match found".

qiime demux emp-paired
--i-seqs
--o-per-sample-sequences
--m-barcodes-file
--m-barcodes-column
--o-error-correction-details

However, when adding these two parameters to the command, the demux step would run successfully. (It had to be both of them, not one or the other)
*--p-rev-comp-barcodes *
--p-rev-comp-mapping-barcodes

  1. What I'm trying to understand is that, why these reverse parameters would make it work? Reversing both the barcode sequences and mapping file barcodes seem to have canceled out the effect.

  2. One colleague of mine recommended trying without these two reverse parameters but adding --p-no-golay-error-correction. The code would also run successfully, and the demux result was more stringent. I have concerns because we are actually using Golay barcodes (100% confident). Therefore, I also hope to understand what is this parameter actually doing. Would it be ok to use it on Golay barcodes if adding it makes the code run properly?

Thank you very much.

2

@sandray1007 Just chime in here, I did some search on this forum and it seems similar questions have been asked and answered (e.g. [Demuxing golay barcodes?]). You have to specify --p-rev-comp-barcodes and --p-rev-comp-mapping-barcodes at the same time, IF you are using the current version of EMP protocol, in which the Golay barcode is on the forward primer.

As explained by thermokarst here Demuxing golay barcodes? - #8 by thermokarst, "With Golay error correction, the orientation matters 100% — the barcode sequences and the mapping sequence both need to be in the 5′ -> 3′ orientation in order for". What this suggests is that the Golay barcodes in the current version of EMP protocol is actually in the wrong orientation: they are the reverse complement to the actual golay barcodes. That is why you have to set --p-rev-comp-barcodes to let QIIME know that the barcode reads need to be reverse-complemented before error correction, and --p-rev-comp-mapping-barcodes to let QIIME know to reverse-complement the barcode in the mapping file before demultiplexing.

If my understanding is correct, then I would think the EMP protocol needs to put a notice that the Golay barcode is not actually the Golay barcode but its reverse complement.