End Repair problem with gDNA library prep (TruSeq)

[Hilary April Smith]

Hi. We've been trying to prepare whole-genome DNA libraries with the TruSeq DNA library sample prep kit. I've been sharing reagents/kits, and had to split my last library prep between an older kit and a newer one. The old kit (not expired) yielded ~0.1-2.4 nM; the newer kit yielded ≥89 nM per sample. I think the problem was either age or multiple freeze/thaw cycles for the End Repair mix. We're trying to determine if it's ok to just re-amplify the libraries with low yield by a couple extra PCR cycles (we already did 10), or if we should re-make the libraries. Our concern is whether (1) the problem -- End Repair as I suspect or something else -- could somehow bias our results and yield to different results between the libraries with the two kits, and (2) if doing >10 PCR cycles could result in undue bias.

Note we're only using 1/2 reagent volume to save $, and starting with only ~300 ng dsDNA (some fresh, some old) due to limiting material. Bioanalyzer traces show ~500 bp for the size for libraries from either kit -- which I assume(?) means that the adapters ligated and we just have low yield. (We finally fine-tuned the BluePippin preps to work, but now this is causing us to ? about 22 libraries).

Any help is greatly appreciated!!!
Best,
Hilary

[ECO]

Why can't you sequence them? Just concerned about the discrepant yield?

[Hilary April Smith]

I think that we will need to do a couple more PCR cycles to have enough, as some of our samples are too low (0.1 nM). We're only multiplexing 7 samples per lane, so normalizing everything to the same concentration won't yield enough to load on the flow cell.

Our bigger concern is whether the varying output from the two kits means that there would be some bias to the results. If it is only a difference in yield, then we can just do PCR and fix it. IF there's some likelihood of bias (eg if the low yield means that we'd have poor coverage -- perhaps losing data from single copy regions if the low yield means we only have things in multiple copies that came through) then we'd need to remake them. We're trying to make sure that if we spend the $ to sequence them that we have reasonable faith in the results, since we're using 4 lanes PE and thus the sequencing costs greatly outweigh library construction costs (if one excludes the time...). Granted there's no way to know for sure without sequencing, but as we're still a bit newer to the TruSeq kits, we're grateful for any insight/suggestions.

[pmiguel]

Hi Hilary,
0.1 nM is around 60 million amplicons/ul. What is your volume? You could speed vap them.

--
Phillip

[Hilary April Smith]

We have ~11 uL left (roughly) after Bioanalyzer and quantification runs; I had halved reaction volumes and did an extra elution at the end to ensure I didn't have any Ampure bead carry-over. We're multiplexing 7 samples per lane before sending them out for sequencing (we don't have a HiSeq at ND). We'd thought to send the samples to BGI from another lab's recommendation, and a quote we received lists desired qty as 2.98 nM concentrations in 15 uL, or a total of at least 0.07 pmol -- so we're a little under. Yet if one can obtain 60 million amplicons/uL from 0.1 nM and given I thought current v3 chemistry yields ~120-180 million reads/lane, perhaps what we have is enough?

Do you think these libraries would still be viable, or that the low yield is likely indicative of some greater flaw? It seems ok to me from the Bioanalyzer traces (peak at ~490/500 bp), but I was nervous when I saw the difference in quantity (~2 orders of magnitude) btw the two kits of reagents, having made all libraries at the same time. It clearly is a reagent issue, as some libraries with the same adapter/DNA type (just a different specimen) fell between the two kits, with low yield from the old kit and high from the new.

Thank you very much for your help and thoughts! Given these samples are hard to come by (African anophelines) I'm trying to take extra care.
Best,
Hilary

[pmiguel]

Hi Hilary,
Your concerns are justified. But sometimes you are working with a sample that can't be replaced, so you go with what you have.

BTW, the efficiency of clustering is not 100% on the HiSeq, so even though your tube probably has 600 million amplicons in it, the standard Illumina denaturation protocol leads to most of the sample being thrown away. BGI probably doesn't want to tinker around with clustering a sample below their specs. So I would say remake the library, if you can. If not, yes do some more PCR.

The thing is that frequently when we squeeze a run out of a sample that doesn't want to give a good library, we end up with high percentage of whatever happens to contaminate the TruSeq kit. Also, depends on what the sample is for. If for de novo transcriptome (discovery) stuff, it may still be useful. If it is a replicate, maybe not.

--
Phillip

[Hilary April Smith]

Thank you very much. I think we will be remaking the libraries; upon further discussion it seems the older kit went through ~6-7 freeze / thaw cycles before we received it, so that probably explains the low yield.