inline barcodes appearing in reverse read of pair

[dkainer]

Hi All,

I have Illumina 101bp paired-end data where the libraries were prepared with custom inline barcodes at the P5 end. So the first 6 bases of the R1 reads contain the barcode sequences for demultiplexing. I used Flexbar 2.5 to demultiplex the reads and it assigned about 85% of them successfully (lower than i would of liked, but ok for now). However, when I look in the R2 reads for each sample I can see that there is a disproportionate number that have the barcode as well.

For example, with sample DK097, the output can be broken down as:

R1: [BARCODE] [ READ SEQUENCE ] 11.0 million reads
R2: [ READ SEQUENCE ] 10.5 million reads
R2: [BARCODE] [ READ SEQUENCE ] 0.5 million reads

What could cause the barcode to be appearing in the reverse read, often right at the start of the read, in up to 5% of the pairs??

cheers
DK

[dpryan]

Maybe the 0.5 million reads have sandwiched adapters, i.e., adapters with adapters on them. This would produce something like:

P5-barcode-read-barcode-P5-P7

I wouldn't expect the rate of that to be so high, but that's the first explanation that comes to my mind at least.

[dkainer]

Hi dpryan,

thanks for the quick reply. So let me try understand this. If we have fragments with:

P5-barcode-read-barcode-P5-P7

When the sequencer does the second read starting from the P7 primer, wouldn't the read contain the P5 adapter at the start before the barcode?

[dpryan]

Unless you already trimmed adapters yes, that's correct.

[dkainer]

no, i can see the barcode at the start of the raw R2 reads prior to trimming.

Let me rename things a little:

P5-barcode---read---barcode*-P5*-P7

Could the ---barcode*-P5*-P7 end of the fragment hybridize to the P5 adapter on the flowcell during the second read phase (instead of the P7 doing it's job), thus causing the R2 read to begin from the P5* sequencing primer?

[SNPsaurus]

Do you have any short fragments so that the first read traverses into the adapters on the other side? That would give some concrete information about the structure of that side. Is this a RAD-type library or something else that uses inline barcodes?

[SNPsaurus]

I was thinking this sounded familiar. Another case?
https://www.jiscmail.ac.uk/cgi-bin/w...NG&F=&S=&P=560

[dkainer]

SNPsaurus, that case looks pretty similar!

The library is whole genome shotgun with P5 barcoded adapters: These adapters are of the form CTTTCCCTACACGACGCTCTTCCGATCT[AGGATG] where those last 6 bases are the barcode.

Here is an R1 (forward) read that is short and traverses through the P7 adapter and Primer 2 (bold). Barcode is blue

AGGATGCACCCGGGAACCTTGATTCACATCCTAGATCGGAAGAGCGGTTCAGCAGGAATGCCGAGACCGATCTCGTATGCCGTCTTCTGCTTGAAAAAAAA


Here is another one which is really just a barcode and then the P7 stuff, then junk

CGACTCAGATCGGAAGAGCGGTTCAGCAGGAATGCCGAGACCGATCTCGTATGCCGTCTTCTGCTTGAAAAAAAAAAAATATTAAAGCAACTACTAGATAG

[nucacidhunter]

I wonder if Q scores for read 2 starting with barcode is lower than normal ones and if the pair maps to reference at expected distance and orientation.