Alternative sizing method for 454 PE libraries?

[CC_seqanswers]

We have had lots of trouble with getting right size of 454 PE libraries. I wonder if one could use gel sizing instead of Ampure beads? Beads sizing always give us inappropriate size or undesired small fragments.

Any thoughts?

[pmiguel]

Sure. We do that sometimes. Ampure is just supposed to be fast. But if you prefer to do an agarose gel size selection, no reason not to.

--
Phillip

[CC_seqanswers]

Thank you, Phillip.

I just ran a PE library on 4-12% TBE polyacrylmide gel from Invitrogen and cut between 300-700 and then elute DNA out of the gel. The size range looked very nice on DNA chip (300-700bp with very only very small percentage of <300 bp) , however, the single stranded DNA looked very different on RNA chip as there was much more small fragments and the peak shifted by 100nt smaller.

Can you suggest why?

btw, to run ss DNA on RNA chip, is the DNA supposed to be heat-denatured as RNA?

Thanks a lot!

gel

Sure. We do that sometimes. Ampure is just supposed to be fast. But if you prefer to do an agarose gel size selection, no reason not to.

--
Phillip

[pmiguel]

Thank you, Phillip.

I just ran a PE library on 4-12% TBE polyacrylmide gel from Invitrogen and cut between 300-700 and then elute DNA out of the gel. The size range looked very nice on DNA chip (300-700bp with very only very small percentage of <300 bp) , however, the single stranded DNA looked very different on RNA chip as there was much more small fragments and the peak shifted by 100nt smaller.

Can you suggest why?

It's complicated.

Double stranded molecules, near as I can tell, run faster than single stranded molecules on Agilent chips. See, for example:

http://seqanswers.com/forums/showthread.php?t=12501

http://seqanswers.com/forums/showthread.php?t=12852

Denaturing a double-stranded DNA sample by heading to 95 oC for 2 minutes and then placing the tube immediately on ice, appears to convert it to ssDNA when run on a picoRNA chip. But obviously if there is a little salt in your sample, or it has some extra time to re-anneal, it might begin to return to a double stranded state.

To make matters more complicated, I think it is common for some fraction of the PCR amplification artifacts (primer-dimers, or whatever) to become complexed with the larger library molecules. Like the adapters anneal to a part of a primer-dimer. The primer-dimer might even "splint" across a couple of real library amplicons, creating extra, higher molecular weight peak(s).

As far as your sample goes, if you want to post the electropherogram here, I would be interested to take a look.

btw, to run ss DNA on RNA chip, is the DNA supposed to be heat-denatured as RNA?

Thanks a lot!

gel

Yes, I think so. Did you heat-denature? What conditions did you use?

--
Phillip

[CC_seqanswers]

I attached two Agilent traces. I marked size range and hope it makes sense to you.

After gel sizing and DNA elution, dsDNA was run on DNA chip; and then ssDNA was isolated based on 454 PE protocol. ssDNA was later run on a RNA chip. It was 70C denatured for 2 minutes and chilled before loaded.

Thanks very much for your help!

It's complicated.

Double stranded molecules, near as I can tell, run faster than single stranded molecules on Agilent chips. See, for example:

http://seqanswers.com/forums/showthread.php?t=12501

http://seqanswers.com/forums/showthread.php?t=12852

Denaturing a double-stranded DNA sample by heading to 95 oC for 2 minutes and then placing the tube immediately on ice, appears to convert it to ssDNA when run on a picoRNA chip. But obviously if there is a little salt in your sample, or it has some extra time to re-anneal, it might begin to return to a double stranded state.

To make matters more complicated, I think it is common for some fraction of the PCR amplification artifacts (primer-dimers, or whatever) to become complexed with the larger library molecules. Like the adapters anneal to a part of a primer-dimer. The primer-dimer might even "splint" across a couple of real library amplicons, creating extra, higher molecular weight peak(s).

As far as your sample goes, if you want to post the electropherogram here, I would be interested to take a look.


Yes, I think so. Did you heat-denature? What conditions did you use?

--
Phillip

[pmiguel]

I attached two Agilent traces. I marked size range and hope it makes sense to you.

After gel sizing and DNA elution, dsDNA was run on DNA chip; and then ssDNA was isolated based on 454 PE protocol. ssDNA was later run on a RNA chip. It was 70C denatured for 2 minutes and chilled before loaded.

Thanks very much for your help!

Seems clear to me that the 128 nt amplicon was complexed with longer amplicons while double stranded. Basically a "trapped" primer dimer. Then you released it when eluting single stranded from the streptavidin beads.

So the 128 nt amplicon has the B-adapter on one side, and not on the other. If the other side has the A-adapter, then it is a real amplicon and will give you some short sequences mixed in with your normal length ones. You could check the molar concentration of the peaks, see if it is likely to cause you issues.

That said, I am just speculating. Could be other reasonably hypotheses as to what is going on. If you sequence it, please post your results. Specifically your read length histogram.

Next time I guess you could run your library on a denaturing PAGE gel if you wanted to exactly size it. As long as the urea would not interfere with your biotin binding step downstream.

Thanks for posting, interesting.

--
Phillip

[CC_seqanswers]

Thanks! It made lots of sense.

Would you think this might be one reason to use beads for sizing for 454 PE protocol? It seems to me that beads won't help with the issue. However, for this certain library, now that it's ssDNA, if I perform a round of bead size selection will it sufficiently remove the small fragments, say use 0.5x beads volume?


I have concerns with urea gel too because of biotin associated ssDNA extraction downstream.

Do you know the primer/adaptor size for 454 PE library? Is it supposed to be 128nt+?

Seems clear to me that the 128 nt amplicon was complexed with longer amplicons while double stranded. Basically a "trapped" primer dimer. Then you released it when eluting single stranded from the streptavidin beads.

So the 128 nt amplicon has the B-adapter on one side, and not on the other. If the other side has the A-adapter, then it is a real amplicon and will give you some short sequences mixed in with your normal length ones. You could check the molar concentration of the peaks, see if it is likely to cause you issues.

That said, I am just speculating. Could be other reasonably hypotheses as to what is going on. If you sequence it, please post your results. Specifically your read length histogram.

Next time I guess you could run your library on a denaturing PAGE gel if you wanted to exactly size it. As long as the urea would not interfere with your biotin binding step downstream.

Thanks for posting, interesting.

--
Phillip

[pmiguel]

Thanks! It made lots of sense.

Would you think this might be one reason to use beads for sizing for 454 PE protocol? It seems to me that beads won't help with the issue. However, for this certain library, now that it's ssDNA, if I perform a round of bead size selection will it sufficiently remove the small fragments, say use 0.5x beads volume?

In principle. However you would need to calibrate your Ampure beads on ssDNA, probably. Otherwise I'm not sure how to translate dsDNA MW cutoffs into ssDNA MW cutoffs.

I have concerns with urea gel too because of biotin associated ssDNA extraction downstream.

Do you know the primer/adaptor size for 454 PE library? Is it supposed to be 128nt+?

The adapters are 30 bp and the internal linker (after recombination) is 41 bp. So, no, it does not fit the size of an adapter dimer.

--
Phillip

[CC_seqanswers]

I was told the same cut off could be used at least in this case. But I am not sure either as it takes some work to validate it.

May I ask a couple of more questions regarding Ampure beads?
1. have you been always be able to produce 454 PE libraries of idea size with calibrated Ampure XP beads? Did it happen that small fragments still hang round ?What's the PE cutoff value worked for you? I have seen a few old protocols using 0.7 for PE and then 0.3 to remove small fragments (I am still trying to figure out how this double-sizing work. Any hits?). Also I have heard ampure XP does not really vary from lot to lot so some labs actually skip validations.

2. based on your experience, what's beads/DNA ratio works well to remove <300bp DNA in the case of 454 libraries? I tried both 0.5x and 0.4 but did not have a chance to validate it.

You are so appreciated!

In principle. However you would need to calibrate your Ampure beads on ssDNA, probably. Otherwise I'm not sure how to translate dsDNA MW cutoffs into ssDNA MW cutoffs.



The adapters are 30 bp and the internal linker (after recombination) is 41 bp. So, no, it does not fit the size of an adapter dimer.

--
Phillip

[pmiguel]

I was told the same cut off could be used at least in this case. But I am not sure either as it takes some work to validate it.

May I ask a couple of more questions regarding Ampure beads?
1. have you been always be able to produce 454 PE libraries of idea size with calibrated Ampure XP beads? Did it happen that small fragments still hang round ?What's the PE cutoff value worked for you? I have seen a few old protocols using 0.7 for PE and then 0.3 to remove small fragments (I am still trying to figure out how this double-sizing work. Any hits?). Also I have heard ampure XP does not really vary from lot to lot so some labs actually skip validations.

We almost always use agarose gels, Pippin prep or Egels to do size selections. So I guess the answer is "no", but we have not really tried very hard.

It is only since we have been making Illumina libraries that I have started to look carefully at those smaller fragments using a pico RNA chip.

By the way, we never do that streptavidin ssDNA step for 454 library construction. Lost all our sample too many times. We just denature the (double stranded) library prior to emPCR. Plus, most of our libraries are Rapid libraries now anyway...

2. based on your experience, what's beads/DNA ratio works well to remove <300bp DNA in the case of 454 libraries? I tried both 0.5x and 0.4 but did not have a chance to validate it.

You are so appreciated!

Recently I have begun to suspect that the Ampure works fine at removing small DNA. The problem is mainly annealed DNA complexes generated during "enrichment" PCR. So it would be nice to have an Ampure technique that would work under strand denaturing conditions. But that may not be possible, given that conditions that denature DNA may also prevent them from precipitating.

--
Phillip

[CC_seqanswers]

Your post is always filled with neat information.

I actually thought of denaturing PE libraries before ePCR like with rapid library. Do not quite understand the necessity of using strepvidin beads. Unless running ssDNA on RNA chip gives a better view on fragment size.

Now another question got brought up. If one denatures PE library before ePCR, does it change the number of molecules/bead for ePCR as one dsDNA molecule will be denatured into two ss DNA molecules?

454 is our new toy.. Guess I got millions of questions.

We almost always use agarose gels, Pippin prep or Egels to do size selections. So I guess the answer is "no", but we have not really tried very hard.

It is only since we have been making Illumina libraries that I have started to look carefully at those smaller fragments using a pico RNA chip.

By the way, we never do that streptavidin ssDNA step for 454 library construction. Lost all our sample too many times. We just denature the (double stranded) library prior to emPCR. Plus, most of our libraries are Rapid libraries now anyway...



Recently I have begun to suspect that the Ampure works fine at removing small DNA. The problem is mainly annealed DNA complexes generated during "enrichment" PCR. So it would be nice to have an Ampure technique that would work under strand denaturing conditions. But that may not be possible, given that conditions that denature DNA may also prevent them from precipitating.

--
Phillip

[pmiguel]

Yes, both strands should amplify.

The SA Beads thing was a way to deal with non-amplicon library molecules. You have 2 adapters, "A" and "B". To amplify during emPCR and have a sequencing priming site on and amplicon, it needs to have "A" on one side and "B" on the other. But luck of the draw will give you "A" on both sides or "B" on both sides 1/2 the time.

So you conjugate one strand of the "B" adapter with biotin. Then allow all your putative library molecules to bind to the beads. The "A"-"A" molecules will not bind at all and are washed away. The "B"-"B" amplicons are bound by both strands to the beads. So, in this scenario, only "A"-"B" amplicons will release a single strand. Perfect.

Well, there are a couple of problems.
(1) Hey, it is an extra step. If something goes wrong, you lose your library or your yield is low.

(2) Not all putative library molecules have adapters on both ends. So "B"-"X" molecules will also release a strand. I always found it grating that the protocol just shrugs its shoulders at this possibility. All it takes is a minor problem with one of the steps of library construction and you could end up with only, say, 10% of fragment ends having an adapter ligated. Then, going into the SA bead binding you would have:


XX 81.00%
XA 9.00%
XB 9.00%
AA 0.25%
BB 0.25%
AB 0.5%

Then after binding and eluting from the SA beads you end up with 18:1 XB to AB. That isn't so bad in and of itself, but I just don't like the idea of those XB's hanging around in the microreactors with the AB's.

Yes, I know the XBs won't amplify. Well, actually they will -- just not exponentially. And it just seems like an opportunity for recombination/chimerics.

Anyway, Roche switched to Y-adapters for their "Rapid" libraries. So it really isn't an issue any more. So this is little more than a history lesson. Next week: how to pour a 0.2 mm urea PAGE sequencing gel. Yeah, forget all this second and third gen, lets go back to the problems that I had back during zero gen. Excuse me, I have to go tell those kids to get off my lawn...

--
Phillip

[CC_seqanswers]

Very impressive information. Thanks again!

Seems the adaptor/ligation design is quite much like SOLiD libraries.Hope 454 will adopt A-tail/Y shape adaptors for PE libraries soon.

We have been lucky with preserving enough ssDNA libraries after ssDNA extraction using SA. The problem is often that size-range is not ideal and at that point there is too little DNA for further sizing.

We have also seen short reads/short peaks after the run although the library seems to be in perfect size, i.e. very small percentage of small fragments. Could not figure out why. Anything funky going on during ePCR?

Yes, both strands should amplify.

The SA Beads thing was a way to deal with non-amplicon library molecules. You have 2 adapters, "A" and "B". To amplify during emPCR and have a sequencing priming site on and amplicon, it needs to have "A" on one side and "B" on the other. But luck of the draw will give you "A" on both sides or "B" on both sides 1/2 the time.

So you conjugate one strand of the "B" adapter with biotin. Then allow all your putative library molecules to bind to the beads. The "A"-"A" molecules will not bind at all and are washed away. The "B"-"B" amplicons are bound by both strands to the beads. So, in this scenario, only "A"-"B" amplicons will release a single strand. Perfect.

Well, there are a couple of problems.
(1) Hey, it is an extra step. If something goes wrong, you lose your library or your yield is low.

(2) Not all putative library molecules have adapters on both ends. So "B"-"X" molecules will also release a strand. I always found it grating that the protocol just shrugs its shoulders at this possibility. All it takes is a minor problem with one of the steps of library construction and you could end up with only, say, 10% of fragment ends having an adapter ligated. Then, going into the SA bead binding you would have:


XX 81.00%
XA 9.00%
XB 9.00%
AA 0.25%
BB 0.25%
AB 0.5%

Then after binding and eluting from the SA beads you end up with 18:1 XB to AB. That isn't so bad in and of itself, but I just don't like the idea of those XB's hanging around in the microreactors with the AB's.

Yes, I know the XBs won't amplify. Well, actually they will -- just not exponentially. And it just seems like an opportunity for recombination/chimerics.

Anyway, Roche switched to Y-adapters for their "Rapid" libraries. So it really isn't an issue any more. So this is little more than a history lesson. Next week: how to pour a 0.2 mm urea PAGE sequencing gel. Yeah, forget all this second and third gen, lets go back to the problems that I had back during zero gen. Excuse me, I have to go tell those kids to get off my lawn...

--
Phillip