TruSeq - ChIP-seq library construction

[pmiguel]

Out of curiosity, anyone know how often klenow would add a base other than "A" as a 3' tail to blunt DNA?

I know you can force Taq to add a "T" by having only dTTP available during the tailing reaction. So that makes me think that polymerases with this capability might (rarely) add bases other than A to a blunt 3' end.

--
Phillip

[kalyankpy]

ChIP-Seq protocol

Hi Ethanol,

Thanks for sharing the protocol. I noticed that you are doing end repair and A-tailing in two steps. We have a shortened protocol that combines end repair and A-tailing for this part of the protocol. You may consider it as it worked perfectly for us (Bseq project)

Sonicated DNA = 16 ul
NEB buff #2 = 2 ul
Klenow fragment=1 ul
dNTP mix = 1 ul (10mM dATP; 1mMdGTP, 1mM dCTP, 1mMdTTP)

Mix and incubate at 30 deg for 20 min (end repair) and 37 deg for 20 min (A-tailing).

I also noticed that Myers lab does it at 37 deg for 30 min in one go! It saves time and resources when there are many libraries to prepare!

Looks like I updated the protocol but not the link.
For now it is here:

http://ethanomics.files.wordpress.com/2011/11/chip_truseq3.pdf

But if I update it, you can still find it on my blog at:

http://ethanomics.wordpress.com/

It is working great. Finishing it off with the Ampure XP purification gets rid of any extra PCR oligos which if exist in excess inhibit binding of the library to the flow cell.

[ETHANol]

Looks like I updated the protocol but not the link.
For now it is here:

http://ethanomics.files.wordpress.com/2011/11/chip_truseq3.pdf

But if I update it, you can still find it on my blog at:

http://ethanomics.wordpress.com/

It is working great. Finishing it off with the Ampure XP purification gets rid of any extra PCR oligos which if exist in excess inhibit binding of the library to the flow cell.

[kalyankpy]

Share the protocol

After searching the web ways to barcode ChIP-seq libraries, it seems like the best option is to add a barcode at the end of the adaptors (Lefrancois et al 2009) but this requires you to make libraries in multiples of four and the invariant T I'm told may cause issues during sequencing. Illumina still has not released TruSeq for ChIP-seq so I put together a protocol which seems like it should work and uses the reagents we have from the previous ChIP-seq protocol. Illumina has released the adapter sequences but I had to take my best guess at the sequence of the PCR primers.

After spending a beautiful summer weekend in the lab dialing this protocol in, I get good amplification and a nice smear in the correct size range with only 11 cycles of PCR (total cycles from first and second PCR steps) and no self-ligated adaptors. So in theory it looks good.

Has anyone tried ChIP-seq with TruSeq primers? What's your opinion of the attached protocol? Do you think it will work or am I about to waste a thousand dollars on sequencing junk?

Any comments or improvements are appreciated.

A slightly updated version of the protocol is here

http://ethanomics.files.wordpress.com/2011/10/chip_truseq.pdf

Attached file removed check the updated version on my blog.

Dear Ethanol,

We also tried a chip-seq with trueseq kit without much success. Can u share the protocol with us!

Kalyan

[Jon_Keats]

Hi Ethan,

I've been meaning to thank you for the CHIP-seq protocol. It seems to work great and everyone loves the idea of doing the pre-QPCR to figure out the optimum number of amplification cycles. I have a feeling this will become standard practice in all our library preps, particularly those that call for a significant number of amplification cycles. I'll try to have the lab post the modifications we made but briefly we followed your methods less:

CHIP is done using the Invitrogen Magnify kit after covaris sonication
We used the NEB mastermix CHIP kit

So far the libraries look great and the Kapa amplification is clearly light years better than the Phusion amplification, so thanks for the suggestion.

We are going to test Ampure versus the Qiagen columns in the near future and I'll try to update the outcome of those tests.

Thanks for sharing

[captainentropy]

FWIW, for library PCR (TruSeq kit) I've recently been using PE 1.0 from the old GAII kits (the sequence of which is readily available) and a homemade one I call TruSeq-R. The sequence is: 5' CAAGCAGAAGACGGCATACGAGAT 3'. It works great for library quantitation too.

[pmiguel]

We pretty much always do a gel-based size selection prior to the enrichment PCR. No adapter dimer if you do that. Actually we generally use fewer cycles of enrichment PCR as well.

BTW that ~85 bp peak is not adapter dimer. That is the PCR primers! Bizarre!

--
Phillip

[jlove]

I switched to using my own (short) P5 and P7 PCR primers and definitely get some dimer now, but I used to get adapter dimer when I was using the PCR components (master mix and primer) that came with the TruSeq kit as well. Did other folks have the same experience? For what it's worth, I started doing two rounds of bead purification after PCR and the dimer problem is greatly reduced, but not eliminated.

[pmiguel]

You would still get linear amplification of the adapter dimer from the short oligo but it should reduce the amount of adapter dimer amplification. Maybe I'll give it a try, but then again - If it's not broken, don't fix it. My protocol works well so why bother.

I agree. Illumina appears to be pushing a "designed by geniuses for use by idiots" approach to protocol design. Probably goes over well with their IP lawyers because it also obfuscates their methodology.

That said, their kits actually do work well. At least if you can pierce the veil they erect around their protocols so you can see the potential failure modes...

--
Phillip

[ETHANol]

advanT, one more thing. There was a BIG typo in the protocol I originally posted. It had the final step of PCR at 98˚C for 1 min. That was meant to be a final extension at 72˚C for one min!!!!! That could definitely cause some of your product to anneal as discussed above.

[ETHANol]

That is interesting and seems to be a good idea. I doubt it would completely kill the amplification of the adapter dimers but it would most likely reduce the efficiency. Although G-T base pairs are reasonable stable I believe, so it might not be as efficient as it seems at the first look. So I'd say at this point, it sounds like it would be a good idea to use the Universal Adapter primer and a short primer corresponding the to the 5' end of the Indexed Adapter in the place of the PCR primer cocktail.

AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATC*T
and
CAAGCAGAAGACGGCATACGA*G

You would still get linear amplification of the adapter dimer from the short oligo but it should reduce the amount of adapter dimer amplification. Maybe I'll give it a try, but then again - If it's not broken, don't fix it. My protocol works well so why bother.

It seems like Illumina may have struggled some with the difficulties associated with removing the large TruSeq adapter dimers from their libraries.

[pmiguel]

Yes, generally should not be a problem. Always irritating to try to guess my average amplicon size on one of those double-humpers though. And, being new to Illumina libraries, I'm paranoid there could be adapter-dimers annealed to the main library molecules.

Speaking of which, did you read any of the patent disclosure for Illumina library construction someone posted in another thread yesterday? I could not make it all the way through -- extremely wordy. But they did belabor the fact that their PCR primers included the 3' T overhang bases as a method of selecting against adapter-dimers during enrichment PCR. Well "belabor" may not be the right term. They made opaque references to a nearly indecipherable diagram in the disclosure over and over again. I guess the idea was not to actually disclose anything, but rather to have the appearance of having disclosed the technique while leaving it a mystery to anyone who could not figure it out for themselves.

But, that was the upshot: to get an adapter dimer you needed to remove the 3' T overhang to allow ligation to occur. But the PCR primers will not amplify such a product because they end with a "T".

But, I was thinking, oligonucleotide synthesis has an incredibly high error rate (compared to any polymerase, anyway.) So there must be at least 1 in 1000 oligos that have an "A" overhang, instead of a "T" overhang.

This was in addition to the phosphorothioate linkage to that overhanging T base -- to protect against exonuclease removal of the base. Sadly T4 polymerase happily cleaves these residues -- and that would be the nuclease most likely in play after the "end polishing" step.

But I digress... Remember the thread where you were trying to ascertain the sequence of the TruSeq enrichment PCR primers? I posted evidence that the length of their PCR primers was something crazy long (like at least the length of the entire 60 nt adapter, if not longer.) Well this would be justification for their using primers that long. To prevent the amplification of adapter dimers.

Part of the reason it seems insane though is that it would require their "PCR Primer Cocktail" (PPC) to actually contain all the index variations (12 or 24 of them) for primer 2.

I hope not! That would lead to rampant index switching!

--
Phillip

[ETHANol]

Yes, you need to include oligos as well (I edited the post for clarity). But I think your idea of running 1 ul on an RNA gel is better. Or just don't worry about it at all. I have sequenced samples with the extra band and the data generated was good.

[pmiguel]

If you really want to get rid of the 450 bp band do one cycle of PCR with your library (98˚ 1 min - 72˚ C 5 min with enzyme and dNTPS) and the big band will magically disappear. But you do not need to do this and it is preferable not to because you just will get one more cycle of PCR bias.

You have to add primer also, right?

By the way, you can just strand denature your sample (95 oC 2 minutes, "snap chill" on ice) just prior to loading it on an RNA bioanalyzer chip for QC. Since the multimers result from ectopic strand annealing, if you assay them single-stranded, you don't see that artifact. At least if they don't reanneal during the assay...

The down-side is that the RNA output result from the 2100 Expert software is much more limited than the DNA output results. Also, if your sample does re-anneal, the dsDNA will run shorter than the ssDNA, whereas the ssDNA amplicons annealed at one or both 60 nt adapter ends will run longer, probably. If you are submitting your sample to a core for QC, you might want to do the denaturation in formamide and submit it in formamide. Well, if you have a good source of pure formamide -- formamide can go bad and then it just messes up many downstream assays. Only for QC though! If submitting for sequencing I don't know what the effect of submitting in formamide instead of water would be.

Oh, I just thought of another down-side to RNA chips -- the size standard sucks! But this may just be because single stranded molecules do not produce the nice tight bands that double stranded molecules do?

--
Phillip

[ETHANol]

avanT, just to be sure I am understanding you correctly you see the 450 bp band after amplification. If so this is from over amplification of your library.

See these threads:

http://seqanswers.com/forums/showthread.php?t=9124

http://seqanswers.com/forums/showthread.php?t=11962

and probably more.

Bottom line is it doesn't matter and that big band is actually 220 bp DNA fragments annealed at the adapters but with non-homologous inserts which form after PCR primers becoming limiting. Your library is fine. Send it for sequencing as is. You could have gotten away with less PCR cycles, but it is not big deal. The DNA will be denatured for cluster generation so it's all ok.

If you really want to get rid of the 450 bp band do one cycle of PCR with your library (98˚ 1 min - 72˚ C 5 min with enzyme, dNTPS and PCR Primer Oligos!!) and the big band will magically disappear. But you do not need to do this and it is preferable not to because you just will get one more cycle of PCR bias.