optimum illumina flowcell cluster density calculation

[kmpphd]

Is there an equation, based on the finished library fragment size, for calculating the molar amount of ssDNA that should be added to one lane of a flowcell?

[HESmith]

(1500 x DNA concentration in ng/ul) / library size in bp = nM concentration

[GW_OK]

I think he's looking more for X pM = Y c/mm^2.

And I think the answer is there is no good answer.

[HESmith]

Maybe kmpphd would like to weigh in: did you get the answer you were seeking?

[kmpphd]

Thank you both for the responses. I should have been more explicit in stating my problem, though GW saw where I was going. I was seeking a way to relate pmoles of library fragments of a certain length to the amount of "real estate" that would eventually be taken up by clusters. For this discussion let's assume a single molecule "cluster," and ignore DNA flexibility. To some approximation, Cluster Area = (# of clusters <in some way related to pmoles of frag>)[3.14(length of frag)^2], right? Based on this rationale, if we then achieve optimal cluster density using X pmoles of a 200 nt fragment, wouldn't we be in the ballpark using 0.25X pmoles of a 400 nt frag?
If equations don't help, as I assume from GW's further comment, what adjustments do you make for different library fragment lengths?

[HESmith]

Okay, I get what you're driving at. I think there are several real-world factors that will skew the theoretical calculation. First, the DNA is flexible, so there will be an effective length that's not linear with actual length. Second, the practical upper limit on cluster density is determined primarily by the ability of the imaging software to resolve individual clusters. Any calculation should take into consideration the software, camera resolution, and detector sensitivity (fluorescence intensity decreases as the area increases). Third, the clusters are distributed randomly instead of evenly, and the probability of cluster overlap is not linear with either cluster number or area.

As a practical matter, we haven't observed much difference between libraries ranging in size from 100bp (some people love to sequence adapter dimers!) to 350bp when loaded at the same molar concentration.

[yaximik]

Cluster density may be influenced by the density of primers anchored to the cell surface. The paper describing chemistry of primer attachment is vague about this and it is possible it this was empirically optimized.