Blue Pippin problem with TruSeq DNA libraries

[Hilary April Smith]

Hi. Our genomics core recently received a new Blue Pippin instrument. Another user and myself each prepared TruSeq DNA whole genome libraries (TruSeq DNA sample Prep kit, v2) and attempted to do a tight size selection with 1.5% cassettes for 450 bp targets. Yet the Bioanalyzer showed our final libraries were only 300 bp. I've since contacted Sage and they noted that the TruSeq Y-shaped adapters alter migration, so that users would select a range of 500-600 bp if they want 450 bp (but also noted this range as "approximate"). Has anyone else done this, esp with the new Blue Pippin that uses internal standards (ie no separately loaded EtBr ladder). I am nervous about losing more libraries testing this out, and given our other alternative (traditional agarose gels) will not be as precise, I really hope we can make the Blue Pippin work. Thank you in advance for any insight.

[pmiguel]

Hi. Our genomics core recently received a new Blue Pippin instrument. Another user and myself each prepared TruSeq DNA whole genome libraries (TruSeq DNA sample Prep kit, v2) and attempted to do a tight size selection with 1.5% cassettes for 450 bp targets. Yet the Bioanalyzer showed our final libraries were only 300 bp.

The Bioanalyzer chip is post amplification?

BTW, I have seen a slide from the Broad that claimed a similar shift in apparent MW caused by ligase continuing to bind to the ligation site. Further they were able to overcome the effect by removing the ligase. (Using proteinase K, IIRC.)

I guess you could just do Pippinprep post amplification. That way you have plenty of library and any Y-adapter/ligase-mediated shifts in apparent MW would be behind you.

--
Phillip

[Hilary April Smith]

Yes; the Bioanalyzer chips were post-amplification.

Interesting that continued ligation could cause a problem. We do use a mix (Illumina's Proprietary "Stop Ligation Buffer") to terminate the ligation, so I don't think that is the problem. Great idea on doing the Blue Pippin prep after the PCR. Do you know if that typically causes any other problems? There are 2 AMPure bead cleanups after the ligation, so I assume that should remove most of the adapters before the PCR ... but I was not sure if the potential for any carry-over in un-ligated adapters could cause a problem for the PCR. (I've meanwhile emailed Sage Science and Illumina to see if they have any recommendations, but I find that sometimes the responses are helpful and others not ... so I'm very grateful for help from an experienced user!).

[pmiguel]

Depends on how tightly bound the Ligase is to an amplicon. During ligation, my understanding is that the ligase is covalently linked with both ends that it is joining. If it remains complexed (with or without stop solution) then it might drag the DNA-ligase complex away from the electrode towards which the DNA is migrating.

Of course this is all just an interpretation of various outcomes the researcher at the Broad saw. Could be something else going on. Actually not even helpful unless you feel like firing up the proteinase K, post-ligation. Even then probably a little risky.

--
Phillip

[Hilary April Smith]

Sage Science noted that using the Pippin after PCR would avoid needing to optimize (which I gather is a bit of trial $ error) the size you tell the program to select. Yet Illumina, when contacted, is saying to not deviate from their protocol. I like your idea of trying the Pippin gel selection after the PCR and we may try that. Thank you again.

[yaximik]

I cannot cite exactly the reference document now, but Illumina's recommendation on use of the gel sizing method (it is on their web site) mentions that selection results can be skewed by the way staining is made (pre-, in gel, or post-stain).

[Hilary April Smith]

Hi. Yes. We have low input DNA (we've been running this with 150 ng; we may be able to increase to ~300 ng this time). Thus we're trying to find an alternative that will give higher yield, and I think the Pippin preps can increase both quantity of yield and precision of the cut. Yet the best I can get from Sage is that an approximate range of 500-600 bp, which sounds like it may need optimization, is often used with the TruSeq DNA kits to get the 450 bp band (ultimate aim: 100 bp PE HiSeq reads). Since we're limited by starting material and already destroyed several samples due to not realizing that we needed some sort of an offset (the ~100 bp offset Sage seems to suggest or ???), I'm trying to see if there's a better route than trial & error to determine the optimal offset (esp. since I also need to produce these quickly; the need for obtaining replacement samples after the first run has already delayed the project more than anticipated...). Illumina seems to keep saying to follow their protocol Exactly ... but given they have both a gel-method and a gel-free method (the latter with Exome enrichment), it would seem some modifications are allowed.

[HESmith]

Hi Hilary,

A previous poster resolved the issue of aberrant migration with four cycles of PCR prior to size selection. (Phillip, do you recall who it may have been? Ethanol, perhaps?) It's enough cycles to eliminate the Y ends, but not so many that your library becomes biased for smaller products (like adapter dimers). The extra step is a hassle, but it works, and, since you'll recover more material, you can reduce the final amplification by the same number of cycles.

Hope that helps,
Harold

[Hilary April Smith]

Dear Harold,
That is great! I'll see if I can backtrack and find the protocol or earlier SeqAnswers thread as I've found this all very informative. I was debating whether to do the PCR before or after (following PMiguel's suggestion) and wouldn't have thought of doing it both times. Thank you so much for your help.
Best,
Hilary

[HESmith]

Hi Hilary,

Found it in ETHANol's ChIP-Seq protocol; the link is http://seqanswers.com/forums/showthread.php?t=13093.

Good luck,
Harold

[Hilary April Smith]

Excellent; thank you very much!

[pmiguel]

Harold and Hilary,
Actually we normally only do about 4 cycles of PCR for a typical DNA TruSeq library. Depends on how things go, but generally 10 cycles is vast overkill and will give you those irritating "bubble product" peaks.

I happened to talk to someone from Sage Science 2 days ago. They said the Y-adapter migration artefact did not happen in the absence of the ethidium bromide in the gel. So, if you have the blue pippen, you should be able to use the ethidium bromide-free cassettes.

But the main reason not to do your amplification prior to size selection is that it will allow some amount of any adapter dimers to anneal to full length library molecules and, hence, avoid size selection. (My observation.) If you don't see an adapter dimer peak prior to size selection, it probably is not an issue for you.

--
Phillip

[HESmith]

But the main reason not to do your amplification prior to size selection is that it will allow some amount of any adapter dimers to anneal to full length library molecules and, hence, avoid size selection.

We haven't observed this phenomenon, even with ChIP-Seq libraries (where adapter dimers are more frequently a problem). Not sure why it's a problem for you; perhaps it's platform-specific? I'd be interested to hear Hilary's results.

-Harold

[pmiguel]

We haven't observed this phenomenon, even with ChIP-Seq libraries (where adapter dimers are more frequently a problem). Not sure why it's a problem for you; perhaps it's platform-specific? I'd be interested to hear Hilary's results.

-Harold

Well, you may not agree with my interpretation of what is happening. But my assay for this phenomenon is seeing adapter dimer sequence in a library that shows no 120 bp peak on a high sensitivity chip after enrichment PCR + Ampure clean up. Since I don't see that peak, I attribute it to ssDNA adapter dimer annealed to library molecules. I posted about it in another thread.

I don't expect it to be a big issue unless during ligation you ended up with a substantial fraction (>5%) of your products being adapter dimers. Given the safe-guards present in the TruSeq kit this is only likely to occur in cases where insert is limiting or an phosphorothioate-bond-competent exonuclease gets carried into the ligation step. (I think T4 DNA polymerase is such an enzyme.)

To tell you the truth, I have mainly seen this phenomenon in RADseq libraries made in another lab -- not using TruSeq kits at all. So its prevalence may be rare with TruSeq libraries. In point of fact, I don't hesitate to do size selection after enrichment PCR. But everything else being equal, doing size selection prior to enrichment PCR seems preferable to me.

(Of course everything else is not equal. The Y-adapter and/or persisting ligase-binding may make doing a pre-enrichment size selection untenable.)

--
Phillip

[HESmith]

Hi Phillip,

I don't dispute your interpretation: if it's in the sequence data, then it must be in the library .

As I said, we have not experienced this problem, but our sample set is limited to TruSeq libraries (albeit a large number of them). Your post suggests that the problem IS platform-specific (i.e., limited to RADseq and/or non-TruSeq libraries), which is good to know.

-Harold