Illumina paired end poor quality in reverse reads

[TKC]

We recently received paired end results sequenced on the HiSeq 2500, and the forward reads are good but the reverse reads are of extremely low quality (.1% pass generous quality filtering thresholds).

I've attached the FastQC results for the forward (R1) and reverse (R2) reads (let me know if you can't access them). My question is- has anyone seen this issue before? I think it is strange that the R1 reads are so good and the R2 reads are so bad. What would you attribute this to? Overclustering on the flowcell? Something wrong during library prep?

Thanks in advance!

[GenoMax]

Let us start with over clustering hypothesis first. Read 1 looks good.

Can you ask the facility that generated this data if something happened to read 2 for flowcell your sample was on? Ask them for the cluster concentration for the flowcell too. Ideally if something went bad with read 2 they should not have released this data.

[TKC]

Thanks for your response- I was thinking that too, that I will just need to contact the facility.

[GenoMax]

While you are waiting for their answer go ahead and do some analysis to ensure that things otherwise look good (i.e. you get expected alignments to the right genome etc).

[ScottC]

You can also ask them for some library statistics... concentration of the final library and fragment size range (usually Bioanalyzer/Tapestation results). Sometimes you will see good first-read results and poor second read results from libraries whose concentrations are quite low and/or fragment sizes are too large (i.e. >1kb for the HiSeq).

[nucacidhunter]

I have not seen FastQC output as bad as this one. There are four possible causes to investigate:

1- Over-clustering as has been suggested in previous posts
2- Degraded Read2 sequencing primer
3- Low diversity in 3' end of library fragments introduced by library prep method
4- Issues with sequencing machine or reagents

[bilyl]

I've seen this several times. In my experience it's almost certainly due to overclustering so you should ask the staff. The reads will look fine in R1, but after the turnaround step the quality gets really bad presumably because the clusters do grow a little bit with the second round of bridge amp.

[lac302]

I've seen this with overclustering on MiSeq runs. My understanding is that it's a combination of resolution between clusters on R2 and depletion of sequencing reagents.

Have you received your cluster density from the sequencing facility?

[SNPsaurus]

I heard of something like this for a custom amplicon project using non-standard adapters/primers. What kind of library is this?