RNA-seq libraries - double peak after size selection

[MLog]

Hi everyone,
I get some weird results when measuring my RNA-seq libraries on Bioanalyzer and now desperately need an advice...
The libraries look like a peak of expected length (300-400 or 400-500) and a higher peak (>1000 bp according to BA) (the image is attached).
I'm using TruSeq RNA v.2 kit, following the protocol (except for using qiagen columns instead of ampure beads and a size-selection step after PCR enrichment).
This is the first time that I'm making RNA-seq libraries but I have some experience with standard DNA libraries - and none of them looked like this. It was always a peak of expected length, sometimes with traces of adapter dimers at 120 bp - but never any peaks higher than expected.
What puzzles me the most is that I got this AFTER size selection. I'm pretty sure that I cut the band about 300-500 bp and certainly not 1500-2000 bp! So this should be something generated from my 300-500 bp fragments. What could it be? A single-stranded DNA? If so, will it affect the sequencing resuts? Does it make sense to try to run these libraries or it would be safer to redo them?
Thank you very much in advance!

[ssing]

Hi there,

I got the exact same pattern you describe (couldn't see the image). See this post for what I found, what others suggested, and to see the end results. Let me know if you'd like more information.

http://seqanswers.com/forums/showthread.php?t=15224

Best of luck!

[MLog]

Hi there,

I got the exact same pattern you describe (couldn't see the image). See this post for what I found, what others suggested, and to see the end results. Let me know if you'd like more information.

http://seqanswers.com/forums/showthread.php?t=15224

Best of luck!

Thanks, ssing! This is a very useful link. In my case the reason is not the beads - just because I don't use them. So I will certainly try heating and cooling.
PS: there was indeed a problem with the attached image but now it should work.

[pbluescript]

It is due to too many cycles of PCR. When you have a high concentration of DNA fragments with complimentary ends, they can hybridize with each other, leading to the longer fragment sizes. Did you amplify each library with the same number of cycles or did you titrate the cycle number?
If you titrate the number of PCR cycles, you can be sure you don't get these effects and introduce minimal PCR bias into your samples.
You can see the gel I have attached to this post for an example of the appearance of these double sized bands as the number of PCR cycles goes from 6-18.

Edit:
The gel is 1.5% agarose post-stained with EtBr. The ladder is an Invitrogen 1kb+ ladder, so the sizes visible are 100,200,300,400,500,650.

[pmiguel]

Hi there,

I got the exact same pattern you describe (couldn't see the image). See this post for what I found, what others suggested, and to see the end results. Let me know if you'd like more information.

http://seqanswers.com/forums/showthread.php?t=15224

Best of luck!

Hi ssing,
Glad that worked for you. However, about the gel you ran after heating/cooling -- was that agarose? If so, did you pre-stain (eg, EtBr in loading dye and/or gel and/or running buffer) or post-stain the gel?

--
Phillip

[pmiguel]

You can see the gel I have attached to this post for an example of the appearance of these double sized bands as the number of PCR cycles goes from 6-18.

Hi pbluescript,
That is a great picture. Could we get info about the sieving matrix (agarose or acrylamide?) and whether you pre or post-stained it? And what the stain was (EtBr, SyBR-safe, SyBr-gold, etc)?

--
Phillip

[pbluescript]

Hi pbluescript,
That is a great picture. Could we get info about the sieving matrix (agarose or acrylamide?) and whether you pre or post-stained it? And what the stain was (EtBr, SyBR-safe, SyBr-gold, etc)?

--
Phillip

Sure. This one was 1.5% agarose post-stained with EtBr. I've actually tried it with 1-2% agarose post-stained with EtBr, Sybr-safe, and Sybr-gold and get identical results. I have also used the Invitrogen E-gels and see the same pattern.

I generally take an aliquot of each library pre-PCR and titrate the number of cycles to decide how many each library needs.

[pmiguel]

Sure. This one was 1.5% agarose post-stained with EtBr. I've actually tried it with 1-2% agarose post-stained with EtBr, Sybr-safe, and Sybr-gold and get identical results. I have also used the Invitrogen E-gels and see the same pattern.

I generally take an aliquot of each library pre-PCR and titrate the number of cycles to decide how many each library needs.

Then I am not sure what to think. I thought most of the double peak phenomenon was the result of the extra sieving resistance BioAnalyzer chip polymer had for ssDNA vs. dsDNA. But you are showing the same phenomenon in agarose -- agarose without run-time staining. Strange.

--
Phillip

[ssing]

Hi Phillip,

It was a pre-stained agarose gel. Thanks!

Sonal

[pbluescript]

Then I am not sure what to think. I thought most of the double peak phenomenon was the result of the extra sieving resistance BioAnalyzer chip polymer had for ssDNA vs. dsDNA. But you are showing the same phenomenon in agarose -- agarose without run-time staining. Strange.

--
Phillip

I always thought it was the complimentary adapter ends from separate molecules hybridizing, increasing the overall length of the fragments. As the concentration goes up, this becomes more likely. I haven't directly tested that though.

[TonyBrooks]

It is due to too many cycles of PCR. When you have a high concentration of DNA fragments with complimentary ends, they can hybridize with each other, leading to the longer fragment sizes. Did you amplify each library with the same number of cycles or did you titrate the cycle number?
If you titrate the number of PCR cycles, you can be sure you don't get these effects and introduce minimal PCR bias into your samples.
You can see the gel I have attached to this post for an example of the appearance of these double sized bands as the number of PCR cycles goes from 6-18.

They should pose no problem during sequencing as you denature before you load onto the cBot.
You could perform and additional PCR cycle with extra primers and mastermix if you want to check. The additional peak should disappear.

[pbluescript]

They should pose no problem during sequencing as you denature before you load onto the cBot.
You could perform and additional PCR cycle with extra primers and mastermix if you want to check. The additional peak should disappear.

I agree. They've never caused me an issue. I use the appearance of the bands more as a guide to the maximum number of PCR cycles I need for a library.

[monkey_SEQ]

Hi All

This will be my first post…also first time attempting the TruSeq protocol (followed exactly - 4ug total RNA, AmPureXP beads, 15cycle PCR amp). I seemed to have gotten exactly the same result as ssing did (http://seqanswers.com/forums/showthread.php?t=15224), strong band at ±300-500bp with a HMW smear going all the way to the wells (1% EtBr pre-stained agarose gel attached). I had a heart attack when I saw this earlier today, but according to the post I have seen, it seems to be a regular result with the Truseq kit. I will try pmiguel’s advice on denaturing the libraries and rerunning the gel just to see if the smear clears up.

It seems then that these libraries are over amplified (if I understand correctly), but what effect will this have on a transcriptome assembling project? I am interested in read counts/ expression levels. Will an over amplified library result in too much of a bias in the results or will it be fine to continue with what I have?

I will still size-select for what its worth. What is a good insert size to aim for when attempting a de novo assembly?

Thanks for any advice

[pbluescript]

Hi All

This will be my first post…also first time attempting the TruSeq protocol (followed exactly - 4ug total RNA, AmPureXP beads, 15cycle PCR amp). I seemed to have gotten exactly the same result as ssing did (http://seqanswers.com/forums/showthread.php?t=15224), strong band at ±300-500bp with a HMW smear going all the way to the wells (1% EtBr pre-stained agarose gel attached). I had a heart attack when I saw this earlier today, but according to the post I have seen, it seems to be a regular result with the Truseq kit. I will try pmiguel’s advice on denaturing the libraries and rerunning the gel just to see if the smear clears up.

It seems then that these libraries are over amplified (if I understand correctly), but what effect will this have on a transcriptome assembling project? I am interested in read counts/ expression levels. Will an over amplified library result in too much of a bias in the results or will it be fine to continue with what I have?

I will still size-select for what its worth. What is a good insert size to aim for when attempting a de novo assembly?

Thanks for any advice

Actually, those look fine to me. The smear is probably due to the high concentration of DNA.

[lilin001]

It is due to too many cycles of PCR. When you have a high concentration of DNA fragments with complimentary ends, they can hybridize with each other, leading to the longer fragment sizes. Did you amplify each library with the same number of cycles or did you titrate the cycle number?
If you titrate the number of PCR cycles, you can be sure you don't get these effects and introduce minimal PCR bias into your samples.
You can see the gel I have attached to this post for an example of the appearance of these double sized bands as the number of PCR cycles goes from 6-18.

Very nice. Could you please provide ladder size? Thanks!