Size selection of ChIP-Seq library

[OptimusBrien]

Hey guys,
I've a question regarding the size selection step in the generation of an illumina library for ChIP-Seq. The particular cells I'm using are especially difficult to sonicate, hence my DNA fragments are approx. 200-1000+ base pairs at the input stage of my ChIP. This means that the normal size selection step would only get a small fraction of my DNA. The average size of my fragments is approx. 500 base pairs and I'd prefer to extend the size selection up the gel a bit to get as much of my DNA as possible. What is the maximum size of fragment that can be efficiently used for library generation? Ideally I'd like also to take a wider size range, say from 350-800 because my fragment smears are relatively diffuse compared to other sonications. Is it possible to do this?
Cheers
Optimus

[DMO]

600bp. You can go above this successfully to about 800 or 1000bp but you need to seriously reduce your loading concentration otherwise your clusters begin to overlap due to the length of your fragments. Also, if you have a very broad range of fragment size the outcoming distribution of reads will not be even to the input. But, that may not be of any concern.

[mudshark]

an option you could consider is to sonicate the reverse crosslinked and de-proteinized DNA before library preparation. works pretty well in our hands using the bioruptor.

[OptimusBrien]

Hi mudshark, thanks for the input! Out of interest how exactly have you done this yourself? What quantity of DNA? In what volume? At what setting? For how long? I've seen a paper in PLOS One where they developed a method using this 2x sonication approach, but they were using a lot of DNA in a rather large volume for the second sonication.
I tried another approach, using a fragmentation buffer, with no success. The whole thing is complicated by the fact I have very little DNA, basically enough for 1 shot at a library prep, and more DNA is not readily available. So anything I try needs to work :/ I have lots of 'input' DNA which I can play around with to try and get something to work so that's what I'm working on at the moment.

[Chipper]

You could try the method published by Bernstein et al in Nature methods last year (Aug. issue I think), they claim to be able to use ~10000 cells. Protocol is similar to Affy chip-chip arrays, with a linear random extension followed by PCR.

[mudshark]

Hi Optimus. We tested the post-purification fragmentation on the input material. As you pointed out it is difficult to evaluate the success on the IP material. We also tried the "fragmentation buffer" that is part of Sigmas WGA kit. That had no effect. I suggest you to optimize the sonication procedure (we use a bioruptor) on as little input material as possible using a bioanalyzer for detection. I guess that the efficiency will depend on the initial size distribution, amount of crosslink, etc therefore it will probably not helpful if I tell you our conditions. Ideally I would have liked to do this using a Covaris instrument. Unfortunately we do not have access to such a machine.

[captainentropy]

an option you could consider is to sonicate the reverse crosslinked and de-proteinized DNA before library preparation. works pretty well in our hands using the bioruptor.

Agreed. We do the same thing. In fact many labs do, and it's been suggested in the original MACS paper for instance. I know the Joint Genome Institute does it as the first step in their library construction.

It's rather easy to do this step. I first sonicate the (already sonicated and reverse crosslinked) input DNA to determine the settings needed for the tubes and volumes I'm using. Then I apply that to the ChIP samples. Keep in mind as the DNA fragments get shorter they becomes more rigid and thus harder to sonicate. Therefore you may need to sonicate for a very long time to get down to the 150bp range. Actually, I have a suggestion. Unless you have a bioruptor holder to use the 0.5 mL tubes just fashion one from what you can and use that in the 1.5 mL holder. Using the smaller tubes allows you to sonicate in smaller volumes and this actually generates better sonication. I actually had a new holder fabricated (just the disc that holds the tubes in place that screws onto the holder) to hold the 0.5 mL tubes. You can buy one from Diagenode but it's like $1500! Mine cost like $400. Anyway, here's an email I got from Covaris regarding sonication with respect to how small the fragments can get (emphasis added):

You are absolutely right about the size limit of DNA fragments that can be achieved with AFA technology. We believe that the limit is 80 bp (fairly narrow peak with apex around 80 bp). To get such a short fragments you need to shear sample for around one hr. The size limit have probably something to do with the mechanism of DNA fragmentation. The acoustic energy involved in shearing DNA to a small fragments creates some inertial cavitation. In this case bubbles collapse, rebound and emit a pressure pulse or microjet that fragments DNA. The number of bubbles collapsing and their energy could cause the simultaneous fragmentation of the same DNA molecule in two different locations. However, the rigidity of DNA strand, which is closely related to the ability to fragment it, is an exponential function of the inverse of DNA strand length - hence the natural size limit.
I haven't done any experiments with crosslinked chromatin to determine size limit, since most people are interested in fragments around 200 bp, but I suspect that it exists and is probably around 100 bp.

If you have any other questions, please do not hesitate to contact me

Regards,

[Xxxx]


[Xxxx Xxxxxxxxxxx] Ph.D.
Covaris, Inc.
14 Gill Street, Unit H
Woburn, MA 01801-1721

[xiaoyuan zi]

hello to all

hello!

I am a new member