Illumina demultiplexed data with inline index sequences (index5+index7)

[Nicholas_Bokulich]

What sequencing platform did you use? See here:

Odd display of demux plot

Thanks for the welcome! Only a few QC steps were performed by the sequencing company, including throwing reads with too much adapter contamination. I don’t think this is the issue. I’ve confirmed that our reads were sequenced with a Novaseq and indeed have only 4 quality scores, 2, 11, 25, 37. I’ve come across this issue raised on GitHub. @benjjneb recommends enforcing monotonicity in the fitted error model by changing the error model matrix. I can comprehend how to do this in R, but I am…