Chopping up a genome

[Bluenoise]

I'm curious as to what ya'll think is the most effective way of taking genomic DNA and chopping it into random fragments of a defined size. Say around 500bp.

 

I'd think ultrasonification but 500 bp is probably to small for that.

 

So maybe a time calibrated Dnase treatment or even a restriction digest.

There's also pcr with random primers.

 

But which of these methods would be the best to represent the whole genome in Random 500bp fragments?

[Bluenoise]

Well decided to go with the DNase treatment.

 

If anyone has a comment don't be too afraid to reply.

[ecoli]

I'm curious as to what ya'll think is the most effective way of taking genomic DNA and chopping it into random fragments of a defined size. Say around 500bp.

 

I'd think ultrasonification but 500 bp is probably to small for that.

 

So maybe a time calibrated Dnase treatment or even a restriction digest.

There's also pcr with random primers.

 

But which of these methods would be the best to represent the whole genome in Random 500bp fragments?

 

Can I ask why you need to do this? And how are you going to check if all your fragments are the same size?

 

I don't know that much about ultrasonification. Your best bet to control bp size might be random primers, but I don't know of any random primers kit that cut to a specific size. DNase treatment is pretty uncontrolled, so you most likely aren't going to get a specific size out of it. A retriction digest only cuts to a specific sequence so, again, you'd have the size consistency problem.

 

What organism are you planning on doing this with and why?

[Bluenoise]

Well I'm basically looking for short DNA sequences that exhibit certain properties. I have desinged a method to assay for this and capture said sequences. Besides that I really can't really go into it in more detail sorry. The organism is a simple bacterium. (I know pretty vauge but I'm held under disclosure conditions so I hope you understand)

 

Well I'm partial to the DNAse treatment since it will not have a bias towards sequences like the random primers or restriction digest would. It is completely non-specific in where it cleaves DNA. Thus I know that if I have a pool with a high enough concentration of fragements they will almost surely cover the genome a few times over in overlapping fragments. I'm afraid if I have a bias towards certain sequences I will loose what I'm looking for.

 

 

Controlling for size is fairly simple though.

First I'd Need to calibrate the DNase reaction to determine the reaction conditions where the mean length of cleavage products falls within the range I'm looking for. After that I just run the cleavage products on a gel and cut out and purify the fragments that run within that size class I desire. Say between 300-500 bp.

[ecoli]

Controlling for size is fairly simple though.

First I'd Need to calibrate the DNase reaction to determine the reaction conditions where the mean length of cleavage products falls within the range I'm looking for. After that I just run the cleavage products on a gel and cut out and purify the fragments that run within that size class I desire. Say between 300-500 bp.

 

 

I didn't know you could control what DNase can cut. In that case, it does seem like your best bet.

[Bluenoise]

Well you can't control what it cuts. But all reactions have a rate associated with them. So you just need to stop it before it's complete, and the DNA will be partially cut. The mean length proportional to the reation time.

[CharonY]

Is there a special reason why it has to be 500bp? Most shotgun libraries have inserts of around 1-4 kb. These can be easily achieved with several shearing methods.

How large is your genome? The problem with DNase (or other enzymatic treatments) is that the resulting libraries are often non-randomly distributed...

[Bluenoise]

Hmmm I don't think I'd agree with that. DNase will produce a random library, as it has no bias to sequence, as long as the DNA is completely purified from associated proteins.

Other enzymes will not ie RENS.

 

I'm sorry you're just going to have to trust me that I need a library of 300-500 bp inserts. 1kb is for to big for my purposes.

[CharonY]

Hmmm I don't think I'd agree with that. DNase will produce a random library, as it has no bias to sequence, as long as the DNA is completely purified from associated proteins.

 

This does not agree with our experience as well as literature, I am afraid. DNAse I preferentially digests in dependence on the sequence specific local structure, especially at sites adjacent to pyrmidines.

See for instance:

Nucleic Acids Res. 2002 December 15; 30(24): e139.

Journal of Molecular Recognition Volume 7, Issue 2 , Pages 65 - 70

 

However you are right in so far that for small fragments enzymatic digests is probably the way to go.

Alternative enzymes migh be CviJ1 (check Sambrook: "Molecular cloning" for more)

[Bluenoise]

CviJ1 sounds interesting. It would be easier to control and produce all blunt ends, but it would definatley have a much higher sequence dependence as well.

 

If your using DNase I in blunt ending conditions (Mn2+, followed by filling of small overhangs). Then wouldn't every site be a site adjacent to a pyrimidine, since there is always a pyrimidine on one of two strands at every site? And thus have no dependence on this?

 

My biggest concern is calibrating the DNase I reacton environment to get semi-reproducible fragmentation.

[CharonY]

Well with DNase I (as with every other partial digestion for that matter) you have to carefully time your digests. More importantly, you have to redo it for every new batch of enzymes (within a manufacturer often but not always it is reproducible, though).

Regardung DNase I specifity: as I mentioned a bit obscurely the interaction of DNase I is dependent on the local structure of your DNA strand and less on the sequence. More precisely it is sensitive to the structure to the minor grove. There is (at least afaik no interaction with the bases per se, which is the reason why DNase I does not recognizes sequences. So it does not simply find a pyrimidine and cleaves there, which would, as you pointed out, yield theoretical cleavages at every site.

Instead the rigidity and depth of the minor grove are the parameters that are recognized by DNaseI. But as you are aware of, these parameters are of course sequence dependent. More importantly though, the groves are unsymmetrical structers as such the position of, in this case, a pyrimidine on one strand is not equal to the presence on the other one. Thus if you have got certain number of pyrimidines (or maybe it was pyrimidine-purine transitions, I forgot) on the "correct" strand you will yield a higher chance of cleavage. This can be a bit of a problem if you have repetitive regions, for instance.

[Bluenoise]

Oh I see what you mean. I guess no matter how you look at it some bias is unavoidable if you want to generate fragements that small.

 

But I think my purposes can tolerate a little DNase I bias.

The genome that I'm fragmenting is only a few megabases in size so even within 1ug I should have tens of billions of copies of it. And I only need a few of every 300bp section represented. So It will be inevitable that I will get that even with a little bias.

 

I want to avoid RENs since no matter how you look at it some sequences will just not contain a cut site within 1kb no matter which ren you use.

[CharonY]

Suppose for that small fragments it is the way to go. First a couple of test runs to estimate digestion rates, then resolve amd and elute appropriate band(s) from gel.

[Dak]

do you know the sequence?

 

how often do you have to do this? repeatedly, or just a few times?

 

am i right in thinking that the objectives are to have multiple 300-500bp fragments that, between them, represent the entire genome? ie, actual randomness of fragment production is not neccesary as long as the above objective is met?

[CharonY]

In theory you are correct, dak (correct me if I am wrong, Bluenoise).

In case of known sequences (for screening purposes) one could indeed try complete digests with a number of enzymes combinations, but with a sequence of a couple megabases it can get a bit tricky. Using singular enzyme the fragments will probably too short for compelte digests (~256 bp for a tetrameric recognition site).

If however, the library is intended for sequencing purposes there is hardly another way than incomplete (hopefully) random digests.

[fibonacci]

i think u can use the nebulizer....

actually it can cut the genome in random size using the pressure from the gas...

[Bluenoise]

It's amusing how this post came back to the top again after so long. Well anyways I might as well give an update.

The DNase I method worked like a charm. I've been able to create a very functional library with amazingly high sequence diversity and genetic coverage. I estimate that 1uL of my 120 uL library has about 10,000 fold genetic coverage of the bacterial genome that I'm working with.

 

Thanks all for your suggestions.