Shoulder in Truseq ChIP sample preparation libraries

[Rosmano]

Hey,

I am a newbie who has been trying to build ChIP libraries using the TruSeq ChIP Sample preparation kit.
After doing protocol optimization we decided to use MinElute columns for pre-PCR purification instead of using Ampure beads XP. After the PCR, we did a Ampure XP clean-up to remove adapter dimers with the recommended ratio of 1:1.

For the Gel size selection, we are using the E-Gel Size-Select Gels to select fragments between 250-300bp of size.

However, now we are having another type of problems, which is the reason of this post. The libraries that we prepared using this method show a shoulder before the peak and the peak is of a much higher weight than the expected one.

Does anyone know the reason for these shoulders?
Do they affect sequencing quality?
Is it possible to eliminate them?

I used the search function and couldn't find a satisfactory answer.

[kwaraska]

The first issue is that the y-adapters of TruSeq do not run true to size on an E-gel. We were off by anywhere from 100 to 300 bases.

We now use a robot for the size selection but I would suggest either find a Pippin Prep instrument or look for double sided SPRI bead for size selection. (It uses the AMPure beads at differing concentrations to remove High MW then low MW)

[Rosmano]

I wouldn't mind to try that, but if possible in the short term I would like to understand what issues may this shoulder on samples 2 and 4 cause.
Why it may be happening and ways to avoid while trying again an E-gel based purification.

I don't have access to a Pippin prep. I could try the double sided SPRI method but is it as tight as gel size selection?

[Rosmano]

So, I suppose that these Bioanalyzer curves don't have any issue?
Could this shoulder be derived from Y-shaped adapters?

I would like ot know if that shoulder may interfere with sequencing, since last time I sequenced libraries that had that shoulder (and small adapter dimer peaks) I had a really poor quality sequencing list.

[pmiguel]

The high MW peak are just bubble products, no? That is, the products that anneal to each other, instead of being replicated during the last cycle of PCR. Since they are a normal duplex only in the adapter region they migrate more slowly than the full double stranded library molecules.

If you google

"bubble product" site:seqanswers.com

you will see probably a hundred posts on this topic. The good news is that bubble products sequence fine. You just have to get the titre right, which can be tricky unless you use qPCR.

--
Phillip

[SLeigh-Brown]

Looking at your bioanalyzer trace, Rosmano, I believe that these are unlikey to be bubble products. We had a lot of trouble with bubble products a few months ago using the Truseq RNA library prep kit, and we were consistently told that bubble products will run at twice the size of your library peak - so you get two separate peaks on your gel, one at ~300bp as you expect, the other at ~600bp.

What you have shown here is a 'shoulder' which is slightly smaller than your expected library size. I often see this on libraries cut by hand from a slab-gel, but it is less common in library size selected with an e-gel.

As long as that shoulder is DNA from your library, your library should sequence without any problem. A mix of small and large DNA products should not influence sequencing quality, as long as quantification is accurate.

One or two words of caution before you proceed:
- Are you pooling libraries with different indexes? If you are, it will be important to check that all of your libraries have the same profile - if not, your relative ratios of each library in the final pool are unlikely to be accurate.
- If you have access to a Nanodrop, I recommend you run a ul of each library and check your 260/230 ratio is low (this will test if you have phenol or other aromatics contamination - http://www.nanodrop.com/Library/T042...ity-Ratios.pdf)
- If you have a library with a complex profile as opposed to a single peak, I would not recommend using the agilent bioanalyzer for quantification. You will have best results using qPCR.
- I suggest you mention the 'shoulder' to your sequencing provider, and send them the trace so that they can take any steps they feel necessary.

Good luck, and well done on your first try - I have seen many worse libraries than this from someone's first ChIPseq prep!

[Rosmano]

Hello!

Thanks everyone to the replies. Well, we continued to troubleshoot the protocol and managed to get rid of that shoulder by doing two things:

1 - Do an Ampure XP beads purification with a 1:1 ratio instead of a MinElute purification before running the e-gel. The Bioanalyzer profile gets slightly smaller shoulders.

2 - Pick DNA of a smaller size. The e-gel size select isn't exactly accurate. When I pick the fragments between 250bp-300bp I usually get libraries with the size distribution that my profiles show.
Afterwards we decided to pick smaller fragments (take DNA when the 200bp marker is in the bottom well) and managed to consistently obtain DNA with a size of 270-285 bps.

The shoulder disappeared in these samples so I am inclined to believe that it was smaller sized fragments that remain in the bottom well. All the libraries that I prepared have a single peak (the last one has a very small adapter dimer peak).

The concentration range of the libraries that I have prepared untill now, as measured on Qubit, are 13,6ng/uL to 25ng/uL with samples 2 4 and 7 amplified with 17 cycles and the last one 18 cycles (that sample had other problems).
To me they seem to have similar profiles.

[Thimble_12]

Hi SLeigh-Brown,

I just noticed you mention having trouble with double peaks in your RNA-Seq libraries - did you ever get to the bottom of what was causing them? We've had similar issues and suspected it was a kit issue, but it's frustrating not really knowing how to cure it!

[pmiguel]

Hi SLeigh-Brown,

I just noticed you mention having trouble with double peaks in your RNA-Seq libraries - did you ever get to the bottom of what was causing them? We've had similar issues and suspected it was a kit issue, but it's frustrating not really knowing how to cure it!

Which type of double peaks? Bubble products? Just do fewer cycles of PCR.

The other issue -- not completely removing the smaller fragments is more complex and depends on your fractionation method.

--
Phillip

[Thimble_12]

We have been getting double peaks, so we suspect it's bubble products.

Hmm, we've tried reducing the cycle number to 10 and are still having trouble. We were concerned that it was a kit issue, but Illumina claim not to have had other users report problems, so I was interested to know how SLeigh-Brown dealt with the issue, because they seem to have had the problem over a similar time-frame.

[pmiguel]

We have been getting double peaks, so we suspect it's bubble products.

Hmm, we've tried reducing the cycle number to 10 and are still having trouble. We were concerned that it was a kit issue, but Illumina claim not to have had other users report problems, so I was interested to know how SLeigh-Brown dealt with the issue, because they seem to have had the problem over a similar time-frame.

We routinely use 8 cycles with the TruSeq RNA kit.

Note that the bubble products normally are not an issue if you titrate your libraries using qPCR prior to clustering and use the left (lower molecular weight) peak to calculate your average amplicon size.

Remember, the presence of bubble products is just an indication that the ratio of product to primers in the PCR reaction has gone above a threshold at which substantial numbers of products anneal to one another before a primer can anneal. In a sense it is an indication that the prior TruSeq reactions worked well.

I think Illumina over-engineered the kit to work over a wide range of input RNA/DNA amounts.

If you really want to avoid bubble products altogether you can do qPCR prior to amplification and then pick a number of cycles that will keep your final concentration in the 2-4 nM range.

--
Phillip