[captainentropy]

That's great you got that to work, TonyBrooks. We tried that too and it didn't work well for us. How big of a range can you elute?

[carmeyeii]

Hello everybody,

I have a VERY basic question about the PCR step for the library preparation but I just can't seem to find the answer in my head nor in the literature. Any help is greatly appreciated.

I understand that the reason for doing an amplification step after the adapter ligation is to enrich your sample for adapter-ligated fragments which is what will actually bind to the flowcell oligos - however, doesn't this produce multiple exact copies of fragments, causing a significant amount of duplicate reads, i.e., the exact same fragment in multiple clusters on the flowcell? I suspect I may be missing out on something very basic, pergaps it is very unlikely that two or more copies of fragment x will actually make it to the hybridization step with the oligos on the flow-cell? I just don't see how having multiple clusters of a given fragment would not be a complete waste of computing power and money...

Thanks a lot!

Carmen

[captainentropy]

That's a good question Carmen. I think if you have a sufficiently complex library the duplication of specific reads is probably low. But it will happen regardless. Keeping the cycle numbers low will likely help minimize this as well. This is why most (all?) peak calling programs have, either as default or an option, a peak "collapsing" feature that will eliminate reads that have identical 5' starting sequence.

Keep in mind if the chromatin you ChIP from was generated using restriction enzyme digestion you will certainly have many more reads with identical 5' sequence. We do this frequently in my lab and in these cases I disable the peak collapsing.

[pmiguel]

Hi Carmen,
I think enrichment PCR is the "sawdust in the steering column" of next gen library construction protocols. That is, it is a bad idea (for the reasons you describe.) But it obscures a host of library issues that otherwise would have to be dealt with directly. Not the least of which is that some characteristic of the TruSeq adapters in TruSeq Prep kits makes them perform extremely poorly for clustering without at least 1 cycle of "enrichment" PCR.

Oh, let's tack on the low percentage conversion of library molecules to clusters. 12 pM is roughly 7.2 million molecules/ul. 120 ul required per HS cBot lane comes to 864 million amplicons. (Or is it twice that, because they began the day double-stranded?) 12 pM might get you 180 million clusters. But no one is going to bother to optimize on that factor because enrichment PCR gives you more amplicons than you are ever going to use anyway.

To be fair, it also works around the issues of dealing with very low concentration solutions of DNA. But the original protocol, that Illumina blessed with their official adoption of it, was a no-enrichment protocol.

--
Phillip

[captainentropy]

Does anyone know if there has been any study that has looked at this issue experimentally? I presume it would be straightforward - sequence a library with and without PCR amplification (during the library construction) and count the number of fragments containing identical 5' starting sequence. I know when I do peak-calling with and without the "collapsing" I typically lose very few reads (suggesting not a lot of identical fragments were created). Plus, those that are made could come from either the library PCR or the cluster generation.

This isn't a peer reviewed paper (it's just a white paper) but it was written by Heng Li, the guy who wrote the Maq alignment program http://lh3lh3.users.sourceforge.net/...ad/PCR-dup.pdf

The basic message is that to reduce PCR duplicates, we should either increase the original pool of distinct molecules before amplication or reduce the number of reads sequenced from the library. Reducing PCR cycles, however, plays little role.

...unless I missed the point of this paper.

[carmeyeii]

But it obscures a host of library issues that otherwise would have to be dealt with directly. Not the least of which is that some characteristic of the TruSeq adapters in TruSeq Prep kits makes them perform extremely poorly for clustering without at least 1 cycle of "enrichment" PCR.

Thanks, Phillip,

I wonder, though, if that has changed since the new Y adapters already contain the flowcell-complementary sequence?

Carmen

[carmeyeii]

Also, for anyone looking, the advantages of having Y-shaped adapters are explained here:

http://seqanswers.com/forums/showthread.php?t=8869

[pmiguel]

Thanks, Phillip,

I wonder, though, if that has changed since the new Y adapters already contain the flowcell-complementary sequence?

Carmen

"new"? This was already the case when we started making Illumina libraries about 2 years ago.

--
Phillip

[babyviva]

Nice.......................................