PCR-free clustering density issues?

[Number6]

Here are some observations for clustering density with PCR-free DNA preps, and I'm wondering if anyone else has observed the same? On several PCR-free preps, we run the samples at our usual concentration, and get ~60-70% of the expected yield. This probably isn't unexpected, but when we increase the concentration on subsequent runs, the cluster density hits a ceiling and then starts decreasing (sometimes dramatically). Here are some numbers from a HiSeq2000 high output flowcell run with three different concentrations of the same library:

Library at 8 pM: 177M total reads/164M PF reads
Same library at 10 pM: 171M total reads/160M PF reads
Same library at 12 pM: 141M total reads/134M PF reads

The cluster plots for the higher concentrations don't look like an overclustered lane (with a lower PF rate). Images look fine....just lower density.

I assume this is because of the library molecules in the mix with only one adaptor ligated on the ends, not both, interfering with cluster formation.

Any ideas/comments? Is this too much DNA going into the ligation step? Any other solutions?

[nucacidhunter]

This probably isn't unexpected, but when we increase the concentration on subsequent runs, the cluster density hits a ceiling and then starts decreasing (sometimes dramatically).

This seems like typical over clustering and your proceeding data also indicates it. But, then you are saying that this is not the case based on other evidence.

I assume this is because of the library molecules in the mix with only one adaptor ligated on the ends, not both, interfering with cluster formation.

That logically is not possible. Fragments with adapter in one end can hybridise to flow cell but they will not be amplified to form clusters, so their interference with clusters would be minimal.

Any ideas/comments? Is this too much DNA going into the ligation step? Any other solutions?)

I wonder what DNA you are referring to? Providing some thumbnail images from those runs of the same library with various input and actual cluster density reported in SAV might give some further ideas to explore. Also was the library for different runs was denatured in one batch or were they denatured separately?

[Number6]

Thanks, nucacidhunter. This is the only time we've set up a titration to look at these, but we've noticed this occasionally with (our) PCR-free preps. Adding more DNA to the flowcell doesn't increase the cluster density and the PF rate doesn't drop like a typical over-clustered lane (like lane 5 in the ppt I've uploaded).

I agree that partially-ligated DNA fragments wouldn't be expected to interfere; I only suggest this because this is the only explanation that makes any sense to us....seems to me there's *something* in this library that's interfering with cluster generation.

[nucacidhunter]

If your cluster number in PhiX control lane and libraries in other three lanes were as you expect, issues with clustering and SBS reagents and the machines would be ruled out. If occasional variation also is only seen for PCR free libraries in the same flow cell that rules out the occasional sequencer issues as well. I have not noticed any difference between these and Nano or other TruSeq libraries clustering. Maybe logging an Illumina technical support case would be good to see what is their explanation.

[NextGenSeq]

PCR free libraries have higher GC content. You need to denature your libraries using fresh NaOH.

Illumina tech support suggested that we use the 1ml aliquots of 1N NaOH from Teknova and use a fresh tube every week. NaOH solutions absorb CO2 and change pH over time.

http://www.teknova.com/1N-SODIUM-HYDROXIDE-1ML-50pk-p/h0224.htm