[pmiguel]

Oh, my mistake, I meant 2x250 vs 2x300, not v3 vs v4.

No, you are right in essence, you just need to decrement the v's. 2x250 is v2 MiSeq chemistry. 2x300 is v3 MiSeq chemistry.

--
Phillip

[Brian Bushnell]

V4 chemistry is only for HiSeq. Philip is discussing results from V3 chemistry for MiSeq. Not sure if you are comparing the two directly?

Oh, my mistake, I meant 2x250 vs 2x300, not v3 vs v4.

[GenoMax]

V4 chemistry is only for HiSeq. Philip is discussing results from V3 chemistry for MiSeq. Not sure if you are comparing the two directly? Are those plots for a MiSeq run or a HiSeq run with V4?

[Brian Bushnell]

Thanks, Phillip.

The median Q-score is above 30 out to ~300 for read 1 and out to ~275 for read 2, which looks quite good to me. And overall it looks substantially better than our recent runs of the V3 chemistry:



Does Illumina actually suggest that the V4 should give Q30+ out to 300bp? If so, the results don't appear bear that up. But otherwise, I think it's an unrealistic expectation. The 2x250 data we are generating is fine, and V4 looks like a substantial improvement, if long reads are useful to your experiments (they are for us). You can always quality-trim to whatever your error threshold is, and 30 seems too high for most purposes.

[pmiguel]

Okay, here is the run from which most of my faith in the 600 cycle (v3) MiSeq kit derives:



Here is the median Qscore version of the same run for Brian:



This run generated 22M pass filter read pairs at a raw cluster density of 1075 K/mm2. 13 billion bases of which 11 billion were Q30 or greater.

This was a couple of genomic DNA libraries (TruSeq NoPCR libs). So, probably the best chance for the technology to shine. Also we didn't push the cluster density to what is specified by Illumina. I think 1200-1400 K/mm2 is specified.

--
Phillip

[pmiguel]

I guessed you missed the post about 14 million total reads.

No I read it.

--
Phillip

[pmiguel]

@Brian the image might not be too clear but it actually shows a boxplot of all the tile date (in this case 38 tiles) for each cycle.

I think Brian wants the median Q20 plot with error bars.

--
Phillip

[WhiteSeal]

@Brian the image might not be too clear but it actually shows a boxplot of all the tile date (in this case 38 tiles) for each cycle.

@genomax data is indeed a de novo run.

I find it interesting that everyone seems to have their own opinion on what a good run is. Might start a new thread for that as I want to find out how everyone determines that. Also do you have your own QC to determine that. But as I said..... New thread for that..

[NextGenSeq]

Eh? The original poster was talking only about the estimated error profile. Not numbers of reads.
Like I said previously, we typically see 2x more data from the 600 cycle kit than the 500 cycle kit.
But, at a minimum, you should be seeing a 36% increase in the number of reads--just from the instrument imaging a larger portion of the flowcell.
If you don't see that, then there is some sort of issue with the libraries, cluster density or machines involved.

--
Phillip

I guessed you missed the post about 14 million total reads.

[Brian Bushnell]

image

I don't think that image is particularly useful. What if all the bases below Q30 were actually Q29, for example? Some sort of box/whisker plot of quality score or expected (or even actual) error rate would be good, but I wouldn't make any decisions about a run based on a %>Q30 plot.

[pmiguel]

Well as shown by the original poster obviously they are not.

Eh? The original poster was talking only about the estimated error profile. Not numbers of reads.
Like I said previously, we typically see 2x more data from the 600 cycle kit than the 500 cycle kit.
But, at a minimum, you should be seeing a 36% increase in the number of reads--just from the instrument imaging a larger portion of the flowcell.
If you don't see that, then there is some sort of issue with the libraries, cluster density or machines involved.

--
Phillip

[NextGenSeq]

The 600 cycle kit uses 38 tiles, whereas the 500 cycle kit uses 28. That should get you 36% more reads from the v3 kit, even if you were getting the same pass filter cluster density.

--
Phillip

Well as shown by the original poster obviously they are not.

[pmiguel]

On the MiSeq we get maybe 24 million versus 20 million for v3. That's not worth it for an extra $400.

The 600 cycle kit uses 38 tiles, whereas the 500 cycle kit uses 28. That should get you 36% more reads from the v3 kit, even if you were getting the same pass filter cluster density.

--
Phillip

[NextGenSeq]

On the MiSeq we get maybe 24 million versus 20 million for v3. That's not worth it for an extra $400.

[pmiguel]

IMHO the v4 chemistry isn't worth the extra money.

40% more in reagent costs gets you roughly twice the data. Generally see >5 billion >=Q30 bases for a 500 cycle run and >10 billion >=Q30 bases for a 600 cycle run.

--
Phillip