miseq clustering higher bottom than top of flow cell

[thermophile]

What can cause different clustering levels on the top vs bottom of the flow cell. I run amplicons, so undercluster generally aim for ~750k clusters. I have a library that I've run twice, both times I'm getting ~700k clusters on the bottom and 200-300k clusters on the top. I looked at the thumbnails since overclustering can lead to poor estimation of cluster density. I really did undercluster-the bottom looks about the same density as I usually get and the top really has about half as much as normal.

library details. 50% bacterial 16s, 30% fungal ITS, 20% phiX. First run 35% PF, 50% phiX aligned, run failed part way through I1. Second run 42% PF, 33% aligned, sample distribution between 16s and ITS is roughly what I expected. Identified top ~40%, bottom ~15%. The 2 runs were on different MiSeq.

[pmiguel]

What can cause different clustering levels on the top vs bottom of the flow cell. I run amplicons, so undercluster generally aim for ~750k clusters. I have a library that I've run twice, both times I'm getting ~700k clusters on the bottom and 200-300k clusters on the top. I looked at the thumbnails since overclustering can lead to poor estimation of cluster density. I really did undercluster-the bottom looks about the same density as I usually get and the top really has about half as much as normal.

Up to this point I'm just thinking "fluidics issue". But you go on to write:

library details. 50% bacterial 16s, 30% fungal ITS, 20% phiX. First run 35% PF, 50% phiX aligned, run failed part way through I1. Second run 42% PF, 33% aligned, sample distribution between 16s and ITS is roughly what I expected. Identified top ~40%, bottom ~15%. The 2 runs were on different MiSeq.

What?
I mean even running the same pool on two MiSeqs could just mean both MiSeq have the same issue -- like a biofilm starting to mess up the fluidics.

But since you did get through R1 the first time, but errored out part way through indexing, I'm surprised. Was it a focus issue? I mean it has to be that the cluster density of i7 producing clusters dropped too low. But that usually seems to happen during the first index cycle.

I'm going to say this isn't one problem it is two. No library issue should cause different cluster densities on the top vs. the bottom of the flowcell--not unless you are doing something crazy like 1/2 filling the sample slot with mineral oil or something else that could cause a top to bottom density gradient.

So, biofilm + some other library issue.

--
Phillip

[thermophile]

The error that halted the first run was (can't remember exact wording) can't find focus without boosting signal beyond threshold. Looking at the thumbnails, the actual focus was fine but there weren't many clusters and it seems the software just gave up after base 6 of the first index.


We do the template line bleach wash between each run. And both machines pass the volume check. Not sure how the volume check works though, is there a time component that would tell me if the pressure gets as high as it's supposed to within the flowcell?

I'm doing a phiX run on the machine where the library failed using a flow cell from the same lot as the ones that look funky. I'll see what that looks like

[pmiguel]

How did your phiX run look?

--
Phillip

[thermophile]

The phiX run failed. Engineer is coming tomorrow. It's possible that it was flow cell related, I used the last of my flowcells from that lot on the phiX run.

I ran another primarily amplicon library on the machine that didn't fail (second machine in my OP). It looks beautiful, ~750k, 75% PF, 80% >Q30. So I'm left with something weird about the library + machine issues OR flow cell issues. I'm going to try mixing the problem library with nextera genomes that I've sequenced before to bump up the base diversity rather than more phiX since they have indexes.

[GW_OK]

It might be a thermal issue.

Edit: But on two different machines?

[thermophile]

thermal issue is certainly on my mind since these are custom amplicons (16S v4). I've asked the engineer to bring a thermocouple to test it. 2 years ago the second machine (the one that's currently working) would pass the thermal ramp test but clearly not hold the correct temperature when we checked with an external thermocouple so had to have it's temp control replaced.

FWIW I use sequencing primers with locking nucleotides to increase the TM since the common v4 primers (Caporasso/Kozich primers) have TM ~62-65

[thermophile]

FAE checked over the first miseq. nothing horribly out of wack. replaced the syringe, realigned the flowcell clamp (it was at the edge of spec which he thought *could* cause some errors). I showed him a bunch of my last runs, which are weird but he can't explain them either. Running another phiX using a new lot of flowcells, it looks fine. So it's still either a flowcell issue or a machine issue. No answers.