HiSeq 2500 v4 chemistry Read 1 quality drop

MikeNice · 03-25-2015, 12:55 PM

We have been seeing a very specific fail mode in about 1 in 10 runs since upgrading to v4 chemistry last year, and we wonder if anyone else is experiencing this or has thoughts to the cause.

When this happens:

the Q30 starts dropping faster than normal, always at about cycle 50,
only the center lanes are affected, 3-6 being most severe, 2 and 7 less so, and 1 and 8 never affected,
always total recovery in the Index and Read 2, with normal Q30 for those reads
accompanied by high pre-phasing.

Illumina suspects it's the "blue bubble" phenomenon that many users see since the upgrade, which is caused by the peltier being at a slightly higher temperature during the reaction (54 degrees instead of 52.5). But that should cause the "blue bubble" pattern on the intensity chart - higher intensity at the bottom of FC due to the cold reagent entering, and then lower intensity as it passes through. We're not seeing that here. Also, our engineer has made temperature adjustments to no avail.

Perhaps v4 chemistry is more sensitive and less robust than previous versions. Has anyone else experienced this?

Thanks!

nucacidhunter · 03-26-2015, 01:36 AM

I have not seen anomaly like this. Just out of interest, are there any differences among good and bad lanes in cluster numbers? Could it be related maybe to a fault in C-Bot?

MikeNice · 03-26-2015, 06:16 AM

Thanks for the reply, nuc. There is no difference between lanes on any early metrics (cluster density, PF, intensities, etc.). We've seen this problem even when we load the same sample in all 8 lanes at the same concentration. We'll have consistent density and PF, then at about cycle 50, the middle lanes start dropping Q30 and pre-phasing jumps. Fluidics always passes tests before and after runs, too.

Jessica D. · 03-26-2015, 06:47 AM

I haven't seen the particular issue you're talking about, either.

None of the usual suspects (solenoid, pump) I'd look at seem to be making sense, considering your fluidics tests pass. Lane dependent issues just always make me think hardware.

Can you share some images off SAV? The Q30 chart would help, but also the intensity heat maps off the flow cell for one or two cycles before the issue through to about five cycles after when it more or less remains steadily bad?

AdriH · 10-20-2015, 07:19 AM

Did you ever resolve your issue? We have seen similar situations with high prefacing in the middle lanes, where lane 3 and 6 are most significant. Also, when looking at the tile images, it seems as if the clusters are becoming more static with each cycle, as if the the next base fails to incorporate. This has only been evident with the HiSeq 2500 v4 run. Also, it seems that a rehyb will resolve the issue, but this is not a permanent fix. If you had a resolution or any feed back it would be much appreciated.

MikeNice · 10-20-2015, 09:38 AM

We never fully confirmed the cause of this, however it was resolved on all but one instrument by having the FSE replace the front manifold. The last instrument was temporarily fixed with a new front manifold, but the problem returned again and again, until we went further upstream and cleaned the VICI selector valve. We haven't seen this issue since. Good luck!

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Quality drop in the Reverse read of paired end library (illumina Hiseq 1000)	anurupa	Illumina/Solexa	2	10-21-2013 11:40 PM
HiSeq 2500	kwaraska	Illumina/Solexa	0	05-30-2013 08:52 AM
Concerns for combining data from HiSeq 2000 and HiSeq 2500	jaaker	Illumina/Solexa	1	02-04-2013 03:56 PM
HiSeq 2500 run, not hopeful on a quality PE-150 quick run	epistatic	Illumina/Solexa	2	10-10-2012 07:47 AM
Illumina second read quality drop	wouter	Illumina/Solexa	4	11-01-2010 06:40 PM

03-25-2015, 12:55 PM	#1
MikeNice Junior Member Location: Washington D.C. Join Date: Mar 2010 Posts: 6	HiSeq 2500 v4 chemistry Read 1 quality drop We have been seeing a very specific fail mode in about 1 in 10 runs since upgrading to v4 chemistry last year, and we wonder if anyone else is experiencing this or has thoughts to the cause. When this happens: the Q30 starts dropping faster than normal, always at about cycle 50, only the center lanes are affected, 3-6 being most severe, 2 and 7 less so, and 1 and 8 never affected, always total recovery in the Index and Read 2, with normal Q30 for those reads accompanied by high pre-phasing. Illumina suspects it's the "blue bubble" phenomenon that many users see since the upgrade, which is caused by the peltier being at a slightly higher temperature during the reaction (54 degrees instead of 52.5). But that should cause the "blue bubble" pattern on the intensity chart - higher intensity at the bottom of FC due to the cold reagent entering, and then lower intensity as it passes through. We're not seeing that here. Also, our engineer has made temperature adjustments to no avail. Perhaps v4 chemistry is more sensitive and less robust than previous versions. Has anyone else experienced this? Thanks!

03-26-2015, 01:36 AM	#2
nucacidhunter Jafar Jabbari Location: Melbourne Join Date: Jan 2013 Posts: 1,238	I have not seen anomaly like this. Just out of interest, are there any differences among good and bad lanes in cluster numbers? Could it be related maybe to a fault in C-Bot?

03-26-2015, 06:16 AM	#3
MikeNice Junior Member Location: Washington D.C. Join Date: Mar 2010 Posts: 6	Thanks for the reply, nuc. There is no difference between lanes on any early metrics (cluster density, PF, intensities, etc.). We've seen this problem even when we load the same sample in all 8 lanes at the same concentration. We'll have consistent density and PF, then at about cycle 50, the middle lanes start dropping Q30 and pre-phasing jumps. Fluidics always passes tests before and after runs, too.

03-26-2015, 06:47 AM	#4
Jessica D. Junior Member Location: Washington, DC Metro Join Date: Feb 2015 Posts: 6	I haven't seen the particular issue you're talking about, either. None of the usual suspects (solenoid, pump) I'd look at seem to be making sense, considering your fluidics tests pass. Lane dependent issues just always make me think hardware. Can you share some images off SAV? The Q30 chart would help, but also the intensity heat maps off the flow cell for one or two cycles before the issue through to about five cycles after when it more or less remains steadily bad?

10-20-2015, 07:19 AM	#5
AdriH Junior Member Location: New York Join Date: Oct 2015 Posts: 1	Did you ever resolve your issue? We have seen similar situations with high prefacing in the middle lanes, where lane 3 and 6 are most significant. Also, when looking at the tile images, it seems as if the clusters are becoming more static with each cycle, as if the the next base fails to incorporate. This has only been evident with the HiSeq 2500 v4 run. Also, it seems that a rehyb will resolve the issue, but this is not a permanent fix. If you had a resolution or any feed back it would be much appreciated.

10-20-2015, 09:38 AM	#6
MikeNice Junior Member Location: Washington D.C. Join Date: Mar 2010 Posts: 6	We never fully confirmed the cause of this, however it was resolved on all but one instrument by having the FSE replace the front manifold. The last instrument was temporarily fixed with a new front manifold, but the problem returned again and again, until we went further upstream and cleaned the VICI selector valve. We haven't seen this issue since. Good luck!