Anyone who’s run Illumina instruments over the years is likely to have noticed how flowcells can have remarkably similar (and occasionally amusing) names. This can create a real headache when looking for a specific run as a single mismatch can cause you to spend some time looking at one run when you thought you were supposed to be looking at something else. Even the BaseSpace search appears to allow mismatches in this key identifier.
This has been an obvious issue their HiSeq platforms for many years but has remained unchanged. It’s only a problem for labs running lots of flowcells because these are generally shipped in a similar order to their manufacturing date. As such it is not uncommon to have two flwocells being run on the same day with different sets of samples on paper and in the LIMs – a recipe for flwocell mixups if ever there was one! And one my team is very careful to avoid!!!
This is what HiSeq flowcells look like (feel free to fill in the gaps):
GA1: GA2: GA2x: HiSeq 2000 2060YABXX (signed by James Watson) HiSeq 2000: C3NBGACXX HiSeq 2500: H9B1YADXX HiSeq 2500: C87KJANXX HiSeq 2500 (rapid): HiSeq 4000: HMJFJBBXX HiSeq X10: HOPETALXX (signed by Jay Flatley) NextSeq: MiniSeq:
NovaSeq looks like it’ll have the same issue – but with many fewer flowcells being run per week due to the vast volumes of data obtainable on Illumina’s newest sequencer:
The problem is worst on MiSeq. Of over 1500 MiSeq runs performed in my lab almost one third of the 5 character flowcell IDs differed by only a single character e.g. A9YV6, A9YVD, A9YVK, A9YVL, A9YVM, A9YVR, A9YVT & A9YVU (a run of 8) or 64D50, 64D53, 64D54, 64D55, 64D5A, 64D5B, 64D5C, 64D5F, 64D5G & 64D5L (a run of 10).
It’d be great if Illumina increased the number of characters, or used these few characters more sensibly, or randomly assigned flowcells IDs rather than simply increment (which is what it looks like they do). Sometimes small things make all the difference.
One of the interesting, but unsubstantiated, things I’ve learnt about Illumina over the years is that they employ a rude word filter to screen flowcell IDs. If you’ve got any interesting ones to share please do (I’ll start a Twitter thread on this).
Fortunately for me they did not seem to catch these little w*****s: C6J0WANXX, C5MCWANXX, C5U6WANXX, C4NYWANXX, C5ECWANXX, C40JWANXX, C4EKWANXX !