1st Cycle Int (PF)

[wouter]

Dear All,

What are 1st Cycle Int (PF) values one could consider to be still acceptable? We have had values going down run after run. We had values around 200 in the first runs we did. These intensities have been going down steadily until we reached values of around 25. At this point error rates went clearly up in the longer reads.

Thanks

[simonandrews]

I'd normally like to see initial intensity values close to 200. I'd be concerned if the values were below 100, and I'm amazed you got any usable data with starting intensities of 25.

[kmcarr]

What Simon said.

If it has steadily been decreasing and is now at 25 you should get your service engineer in as soon as possible.

[wouter]

The mode scramble was replaced in the meantime.
The problem is that the service engineer don't want to give a lower limit for the 1st Cycle Int (PF) at which one should consider it to low.
What kind of values do other people see?

[kmcarr]

We've had the mode scrambler replaced on ours too. We've also had to replace the green laser, the focus laser, the stage as well as a couple of other things I can't remember right now.

I have also had similar experiences with Illumina representatives not giving direct answers to (what I think) are simple questions. Like you they would not define a lower limit for acceptable first base intensity. I'm starting to get the impression that Illumina does not have as good a handle on their technology as they thought they did and are now trying to cover until they learn more themselves.

As for what kind of 1st base intensities we see it can range anywhere from 100-200. If it goes below 100 I feel uneasy. I would really like them to be consistently ~200.

[basickler]

(Disclaimer, I work at Illumina) The reason for the confusion is that the intensity varies quite a bit depending on the model of the GA (GAIIx has a much lower starting intensity than GAII), how long the exposure is set, exactly how the machine was calibrated, the density of the flowcell, the recipe, and the analysis software to name a few. Because of all this, I think technicians don't like to jump to conclusions as many issues overlap. I generally see GAIIs start around 200 and GAIIx usually starts around 100 though this varies by machine. Though I have seen runs that were very good at an intensity of 50 as well. The key is to notice trends in a particular machine, if you see it start to drop consistently it's indicative of a problem.

It's also true that we're still learning quite a bit about the technology and Illumina tends to move things from RnD to production very fast. This gives you the advantage of the crazy leaps in yield we've been seeing but unfortunately things aren't quite as solid as a release that tests for half a year or more before release.

Brad

[wouter]

Intensities drop slowly during the successive cycles. When you do a PE, the second read typically starts at 60% of the first read intensity and it will drop also during successive cycles. When you do PE, 76 bp reads my impression is that an 1st Cycle Int of 50 will make that the intensity will probably drop too low and error rates in the later cycles will increase.
How many cycles where these runs that stared at 50 and where very good?

[basickler]

Actually they were very long runs with a few knobs on the GA cranked up to 11. But you're right, it would have been better had it been higher. The point I wanted to make is that the intensity is very machine and analysis dependent. You're right in saying the intensities should be constant across the machine using the same analysis using a known good sample.

[simonandrews]

Whilst we're on this topic, we had an odd occurrence in a recent run. All of our lanes showed low initial intensities (around 50 or so), but the first base report showed initial intensities of ~200. Despite this the data quality from the run seemed very good with low error rates in PhiX.

Any idea why this sort of discrepancy might suddenly show up. We've not seen this before on our machines, and as far as we're aware we haven't changed anything to cause this. We have an engineer coming to look at the machine, but wondered if anyone else had seen anything similar?