Your optimum cluster density on a HiSeq?

[pmiguel]

Wanted to revisit this topic.

What raw cluster density do you shoot for?

What concentration do you load your denatured templates to achieve this density?

(Also which instrument do you have.) I am thinking High output chemistry, but if you want give your rapid chemistry results, feel free.

By the way, we are seeing TruSeq no-amp DNA libraries cluster at much lower density at a given concentration (qPCR-SYBR green) than we see for other libraries.

For most libraries we do 15 pM and usually end up in the 800-1000 kClusters/mm^2 range. But yesterday we clustered 6 lanes of no-amp libraries. 21-22pM gave us 841-915Kclusters/mm^2.

--
Phillip

[TonyBrooks]

We still aim for the Illumina recommended 850k/mm2, but to be honest if we get a little bit higher then we don't really see any drastic impact on %>Q30 (but that depends a little bit on the library).
I base all coverage calculations on getting 180m reads PF from a lane. We tend to reach that with a density of ~700k. We've also had lanes with over 1000k/mm2 and the data was pretty good, approaching 220m reads PF with ~90% >Q30.
To quantify we use a combination of Bioanalyser and Qubit to pool samples. The final pool is rerun on the Bioanalyser and Kapa qPCR'd. We find we still need to tweak running concentration using knowledge of how the library was prepared. RNA-Seq libraries, we run at 12pM. DNA libraries (specifically ones that use a double SPRI size selection) are better at 11pM. I think the right-hand skew we see in RNA-Seq libraries causes us to over estimate molarity.
We haven't done PCR free libraries yet so I can't comment - no one here has that amount of DNA spare :-)
I also haven't done enough Rapid Runs to notice any trend, suffice to say we've had really variable cluster densities.
FTR we have a 2500 upgraded from a 1000 in January of this year.

[pmiguel]

Hi Tony,
Thanks for the info.
I think we use a very similar procedure to yours for titreing. Athough for us 15 pM seems to give the best result.

(By the way, for anyone still using Illumina phiX libraries to titre their own libraries: BEWARE! The batch to batch variation is high -- not at all uncommon to see them at 7nM instead of the spec. 10nM!)

For rapid runs so far we always just use 12 pM -- and yes, we have seen quite variable results. Although I have the vague sense that the majority of the variance is library-type specific.