Entering edit mode
9.3 years ago
arnstrm
★
1.9k
I have a question about the miseq sequencing. So we performed single-end sequencing using same forward barcode and a different reverse barcodes. Since this is single end, while de mutliplexing we see reads in both direction (50:50). However, some of the samples have un-equal distribution (too many reads in one direction than the other). Is there any particular reason for this?
Thanks for any answers
Did you check the quality along the reads? It is usual for the quality to drop towards the end, this could potentially result in different recall rates for the barcodes for each side. I also seen, on rare occasions systematic errors. For example, I had a dataset where 25% of the reads had N at positions 2 and 3, checking with the latest FastQC I could see these N were spatially clustered.
Thanks, that makes a lot of sense. Also, while demutiplexing we allowed a single base mismatch, causing cross-matches between samples.
Did you design the barcodes yourselves? I think (never done it, just from reading) it is common practice to design barcodes with a hamming distance of two between any two barcodes, to minimize collisions due to sequencing errors. This would also minimize collisions even allowing a single-base mismatch.
Yes, but in our defense, we had to cram as many samples as we could, so we had no choice but to use all possible barcodes (5 bases)! (BTW, this a non-conventional NGS application and we never cared about the depth).
Hey,
while this is true that Illumina has (significant) drops in quality at the end of (especially long) reads. Cameras and flowcell table get recalibrated before read out of the barcodes. This is why you normally see a rise in quality when sequencing the barcodes. I would guess that the barcode problem (hamming distance not large enough) is one of the reason you see what you see! ;) Also if you use the newest version of chemistry (i can only talk about HiSeq2500 v2 - rapid run) quality really stays on top for almost all of the read...