Index read Q30 problems on NextSeq

[Vesperholly2]

Hello, I'm running a singal read dual indexed library on NextSeq, with a custom primer for index 2. While my general run metrics looked fine (clustering and pass filter rates ok) and my Read 1 Q30 was great (91%), both of my index reads crashed out with Q30 values in the 60s. Looking at the flowcell %>Q30 on SAV, once we hit the index reads most of the tiles start coming up grey. Obviously, we didn't get many reads assigned to samples, but I've never seen this before. Any ideas on why this happened, and how to remedy in the future?
Thanks!

[nucacidhunter]

Four possible reasons:

1- Chemistry or mechanical fault, more likely causing failure in stripping read1
2- Sequence of index starting with GG
3- Faults in adapter or primer design or synthesis where index primer does not bind properly, especially non-complementary 3’ base
4- Interaction between Illumina and custom primers

[Vesperholly2]

Hmmm, I wonder about faults in synthesis or chem/mech glitch. The adaptor/primer setup has always worked, so I'm not so curious about that, and we have many different indices in here that all tanked in unison. I wonder about how to best check that beyond simply running it again? If there was a glitch would it show up in the run logs, I wonder, or do those simply collect pass/fail data? No one likes wasting those kits....
Thanks so much for your help!

[nucacidhunter]

Calling Illumina tech support can help at least to rule out some possibilities.

[GenoMax]

Get illumina tech support to take a look at this run. They can look at the instrument remotely (if your's is accessible, they will ask for certain log files) and determine if this problem is reagent/hardware related. In that case you will get a free replacement kit or other additional information to address the problem.

[HESmith]

Another possibility is over-clustering. We've seen the same phenomenon (good reads, failed indices) at high cluster densities.

All of nucacidhunter's explanations would yield index images with no clusters. Check the thumbnails to see if that's the case or not.