Cluster density on a HiSeq2500

[pmiguel]

We are in the midst of our HiSeq2500 upgrade. First one for our intrepid FSE. Still some kinks in the process -- like one of the rear manifolds had an incomplete channel linking it to one of the lanes. Also one of the syringes of the new pump is working at 50% capacity. Still wanted to get test run on the instrument for the weekend, though, so the FSE left the old rear manifold in and we just left the bum pump working at 15/16th of its normal capacity.

I did the denaturation, etc. of phiX myself. Actually Illumina neglected to send any phiX with the install kit, yet called for burning 10 ul of it for the test run. Ack! They wanted to do the denaturation at 10 nM lib/0.1 N NaOH.

I did not even have 10 ul of 10 nM phiX. Could have used another library, but Illumina is doing a goofy 50 nt forward, 100 nt reverse read run. I thought dealing with weird data like that was more trouble that it was worth. So just took 2 ul of 10 nM phiX lib, diluted to 2 nM. The resulting 10 ul was denatured for 5' after the addition of 10 ul of 0.1 M NaOH. Followed this with a dilution to 1 ml with HT1. Then took 500 ul of that and diluted again with 500 ul of HT1. This gave the desired 10 pM denatured phiX lib, albeit with a 5x higher final NaOH concentration than it would have had using the protocol given the FSE.

The upside of this is, that for the HiSeq we generally denature starting with 2nM libs -- so it would actually calibrate our cluster density/lib conc. for us.

Okay, also the amount of denatured library used per lane is much higher than the 120 ul specified for cBot/HiSeq2000: 400 ul, instead of 120.

You just put 400 ul of your denatured library in a 1.5 ml eppendorf and stick the open tube in the fanged loading area -- one tube for each flow cell, and start the run. I think many features are turned off when the instrument is run in this "install mode" run. For instance, though we specified that the run be saved to our network share, the instrument ignored this and saved the runs locally. No big deal, the run worked fine, just could not monitor it remotely.



So looks like at 10 nM we are getting just over 850 Kclusters/mm^2. The bum pump on one flow cell, and the old-fashioned rear manifold on the the other seem to have caused little, if any difference between the two.

Argh! Looks like the instrument only collects data for a single swath/lane in calibration mode! Oh well, maybe we can get a real run on today...

--
Phillip

[SeqOps]

Did you ever determine if 32 tiles is all you get with the current 2500 configs?Just started our validation run. Pretty sure they are going to do something similar to what they did with the MiSeq v2 upgrade: We'll double your output by imaging both surfaces instead of just one....because we forgot then remembered that HiSeqs image both surfaces. Only this time it'll be we'll triple your output by imaging all the tiles that are actually there......then again it could just be that a single optic path instead of multiple was chosen for speed sake.

[pmiguel]

I think this is what should happen:

Validation run only 1 swath per lane (=16 tiles top surface + 16 tiles bottom surface or 32 tiles.) For a non-validation run, Rapid chemistry, you get 2 swaths per lane. That would be 64 tiles/lane.

The (old) standard chemistry does 3 swaths per lane = 96 tiles/lane.

--
Phillip