Amplicon library quantification discorance

[gregrds]

Hello!

I am hoping to get some advice/insight about an issue I have run into a few time with quantifying amplicon libraries for running on MiniSeq and NextSeq machine.

I prep my libraries (~400bp) with PCR and then do QC as follows:
- pool samples, qubit
- use qubit concentrations to dilute down to 2nM
- NEB NextQuant qPCR kit with melt curve
- redilute down to 2nM if needed, repeat qPCR to confirm dilutions if I'm extra suspicious

The library concentration of the amplicons I'm currently working with are always drastically underestimated by the qubit. As a result, my 2nM dilution is usually closer to ~5-10nM based on the qPCR results. I run the qPCR in triplicates and the results are always great (sample and control triplicates have close to identical cq values and melt curve is clean). I usually go with the qPCR concentration, but that has led to under-clustering in runs without PhiX. As a note, diversity shouldn't be a big issue with these libraries. Regardless, this has been an issue for over a year and does not seem to come up with other amplicons/libraries.

To try and understand this, I worked with an Illumina "library expert" person and we went through the following theories:
1. bubble products from over-amplification are leading to qubit underestimation. Seems plausible, but there isn't any evidence of bubble products when I've sent libraries to the fragment analyzer. I'd also assume that if the bubble products are to blame, then the qPCR should be trusted which doesn't explain the under-clustering.
2. some sort of structural anomaly with this specific amplicon makes it cluster inefficiently. To address this I've just started using a heat denaturation step prior to loading the cartridge. This doesn't fully explain the discordance between qubit and qPCR values though.

By spiking in 20-50% PhiX and adding a heat denaturation step prior to loading, the libraries are now sequenced beautifully. But I'm still at a loss for why my qubit measurements underestimate so much. I'd love to hear if anyone has had a similar experience or has some idea of what could be causing this library weirdness. Happy to provide more info if needed, thanks!

[Ben3]

Hello gregrds

I used to have the same problem with the NextSeq. If I quantified using qPCR then the libraries were drastically different values than my values I calculated from Qubit (concentration) and Tapestation (size). I tried many times to figure out what was going on, but never truly understood it. I just decided that I would find the quantification method that works for me and seemed consistent with the loading guidelines from Illumina.

I also wondered whether the qPCR kits were overestimating due to any additional adapters in the mix. I always did more clean-ups than the protocols called for to account for this, but I don't have a great explanation to this problem.

[gregrds]

Thanks for your reply Ben3

Do you mind if I ask what kind of libraries you were working with when you had this issue?

I had the same thought regarding additional adapters causing qPCR to overestimate. I'd think that if that was an issue you'd be able to see a hint of it in either a melt curve or fragment analyzer though, right?

[Ben3]

gregrds yeah I think you're right. You should be able to pick up on something like this on a melt curve, but it's so strange I didn't have a better explanation. I knew another lab that had the same issue and just gave up on the qPCR because they had better results quantifying the other way.

I had this issue mostly with the TruSeq Total RNA Stranded kits. I always got great results on my runs but I could never understand the differences in quantification. I hate when things aren't consistent.

[gregrds]

Ben3 no one seems to know how to explain it. Definitely stumped a few Illumina employees too when I tried to figure it out with them a while back. Labmates tried to help at first but when they couldn't find the issue they chalked it up to me being incapable of correct library prep. Oh the joys of being a woman in a male dominated lab.

Ah gotcha, so I'm assuming your libraries were pretty diverse. Agreed about the inconsistency - it's frustrating having to doubt what people consider the "gold standard" of pre-sequencing quantification. Perhaps we will never know

[Ben3]

gregrds Well I'm sorry about your lab mates. Don't listen to them and know that you're definitely doing everything right. This has been an issue for several labs, so clearly we can't all be doing the same thing wrong.

If you ever gather more insights let me know because I still want to understand why this happens. Best of luck with your lab. I hope it gets better!

[gregrds]

Ben3 thank you, and yeah let me know if you find any more answers as well!