PCR creating very high molecular weight DNA?

[Rocketknight]

Hi all, I've run into a minor problem doing Illumina sample prep + exome enrichment. Everything worked, except when I ran the samples on a Bioanalyzer at the end, I got a peak way out at like 3kb, in addition to the peak I wanted (300-400bp). There's also a tiny hump at about ~650bp, but it's small enough that I'm just going to ignore it.

I'm using Kapa Library Amplification Kits for the first time, so I presume I'm just not optimizing the conditions correctly. I don't think it's contamination, as it's occurring in all of my samples and it's getting amplified in PCR, so it must have adapter sequences in it somewhere, and no-one else in the lab was handling sequencing libraries at the time. Can anyone suggest what might have caused this, and what the high weight DNA could be? Also, what should I do about it? Ignore it, size-select it out, or ditch all my samples and start over?

[ECO]

I'm about 95% sure that peak is composed of library fragments that have annealed to one another via the adapter seqeunces and made a giant interconnected net that runs large. Remember that the DNA bioanalyzer chemostry is non-denaturing and many different artifacts have been observed/discussed here...it's observed mobility is very sensitive to secondary structure.

I'd quantitate with qPCR and sequence that bad boy! (but probably worry that it's been overamplified).

pmiguel incoming in 3...2...

[Rocketknight]

Ack, I cut the cycles from 10 in the standard Illumina protocol to 7 when I switched to Kapa and I still got overamplification? Bah, I'll go down another one or two next time. Thanks though, that explanation makes sense! Are you sure these giant doom-fragments won't interfere with sequencing?

[ECO]

Well not absolutely sure it's overamp. Just that we never see it in our libraries <16 cycles.

I'm not sure of anything with other people's samples.

[pmiguel]

I'm about 95% sure that peak is composed of library fragments that have annealed to one another via the adapter seqeunces and made a giant interconnected net that runs large. Remember that the DNA bioanalyzer chemostry is non-denaturing and many different artifacts have been observed/discussed here...it's observed mobility is very sensitive to secondary structure.

I'd quantitate with qPCR and sequence that bad boy! (but probably worry that it's been overamplified).

pmiguel incoming in 3...2...

Arrgh! Missed my cue. Ah well, not really much to add. Well, one possibility occurs to me.
Rocketknight,
One other source of extraneous peaks on a bioanalyzer chip: seems like if you massively overload one lane, your sample can end up showing up (at the wrong size) in later lanes. But I usually only see this with a high concentration sample loaded on a high sensitivity chip. Which you are clearly not using above.

I guess I could offer the obligatory link to this post. But that really needs to get re-run some time at more reasonable concentrations...

--
Phillip

[Eric@Kapa]

I would agree with ECO and pmiguel -- the high MW smears are most likely to be artefacts due to mis-annealed DNA. This is most often due to "over-amplification", but it could also result from:

-- Finishing the PCR with a denaturation step (followed by relatively rapid cooling), or
-- Using lower than intended/normal concentrations of primers in the amplification.

Both of these situations are analogous to over-amplification, in that you will end up with a diverse population of dsDNA library fragments that become denatured, with incomplete subsequent primer annealing/extension. Because of the high diversity of fragments (high "Cot value"), they are unlikely to re-anneal to their complementary strands, and instead form a mess of partially dsDNA. Of course, adapter sequences are likely to compound and complicate this situation.

While we whole-heartedly encourage the community to discuss such issues here and in other open forums, please don't hesitate to contact [email protected] if you have any further questions and/or feedback.

--
Eric

[NewScientist]

Hi all,

It seems I encounter the same problem as Rocketknight now doing my Illumina library for MySeq. I am trying to produce metagenomes of microbial communities. When I run my library on High sensitivity chip i have an expected peak at 320bp, and another larger and higher at 900bp. More precisely, the peak at 900bp is more like a high MW smear with the highest quantity for 700 to 989bp size.

I tried to do size selection three times on my library to get rid of these large fragments without any success. I tried to denaturate my library (10 min at 96°C, then 15 min at room temperature), no results.

But when I ran my library on a 7500 chip, I just got only one peak at 370 bp and no more peak at 900bp.

I was thinking about doing a cut on Caliper after 400bp but as I see only one peak on DNA 7500, I am now wondering if it is a good idea...

What do you thing about it ?
Shall I start over or do you think it is possible to sequence it ?

Many thanks,

Sarah

[NewScientist]

chips

For information, here are in attachment the HS and 7500 chips of my library.

Sarah

[Rocketknight]

From experience, it's totally okay to sequence these libraries. As ECO suggested, it's almost certainly adapter sequences annealing to each other. When you see this it usually means you did too many PCR cycles, but the library should still sequence fine.

I have no idea why you're only seeing the high-weight fragments on the HS and not on the 7500, though.

[NewScientist]

Thank you very much for your feedback Rocketnight. If we do sequencing I will give some news about it.

Cheers,

Sarah

[huguesparri]

To Newscientist:

We observed the same agilent profile on some of our ChIpseq libraries. The strange thing is that we perform size selection on a gel after the PCR step.
We tried to denature the libraries and measure them again on our Bioanalyzer (using RNA Chip as the material was single stranded after denaturation) and the highest peak was not as high as previously whereas we saw a lot more matrial in the smaller (and correct) one.

We made some investigation and end up with the conclusion that the QG buffer used in the Qiagen gel extraction kit may be the cause of that strange behaviour of our libraries.
And as a matter of fact, we never have this pattern when we use e-gel to size select our material...

[NewScientist]

Thank you for your returns huguesparri. Somebody's already told me about that problem on HS chip with Qiagen gel extraction kit sized libraries.
They had that kind of profile with PCR products purified on gel and decided to sequence it anyway with 454. They got good sequences, but also many small sequences (200bp or less).
On my side, I tried to do my libraries (Illumina) with 5 cycles less and it is pretty much better.
I still get this second peak (now around 450bp) when I run the pure library on HS, but
I dilute 5folds my library and run it on HS chip I only get a single peak (the expected one).
I assume I should reduce the PCR cycles by one or two cycles more.