miseq beginning of r2 very poor quailty

[thermophile]

I'm running amplicons (v4 16s, dual index, custom sequencing primers, Kozich/Schloss primers). I'd been having good luck so was dialing up my clustering density and dialing down my phiX. And then everything fell apart. I have one project that has very low diversity (microbial community wise) that I don't want to run by itself. So my last successful run and this current run that keep failing are both made up of 25% low diversity project samples and 75% regular diversity samples (which are still low diversity as far as sequencing goes).

Here's the q30 heatmap for the successful run:


Now the unsuccessful. This is the second attempt at this exact pool, the first had similar cluster density and phiX as the successful run but R2 was worthless (lots of Ns and 0% phiX aligning in R2) so my FAS wanted me to try rerunning it with lower density and much more phiX.


R1, I1, and I2 are all great. The indexing looks good (I'm multiplexing 120 samples and recovering 115 which is well within my expectations). R2 is bad. And no phiX is aligned. The lack of phiX and the indexing make me pretty confident it's not my library. But I'm at a loss what else it could be. My FAS is kicking this up to second level support, but I'd love to hear if anyone here has had this happen.

[GenoMax]

Is this a 2x300 bp run? I am sure you have seen this thread.

What seems to be odd is that R2 for phiX does not align at all (I assume R1 is aligning ok?). This sounds like you sequencer may be misbehaving/operating outside normal parameters. Have there been other runs on this instrument that were successful? Have you tried a new phiX tube?

[nucacidhunter]

One possible cause of poor R2 could be low diversity of R2 region in comparison to R1 which would be different for different pools of input sample libraries explaining inconsistent run results. If it is not SBS kit or hardware issue you would have to lower density further down to 400-500 K and 20% PhiX for a decent R2 quality. By the way can you post an image of Data By Cycle %Base plot?

[thermophile]

this is a 2x250 we tried the v3 kit twice with poor results so have stuck with v2. Geno-I'd seen that thread in the summer but apparently stopped following it, thanks. I was alerted by Illumina in June about the problem but our runs then were generally bad, I didn't realize that others were having the same 1st run fine, 2nd run very bad issue.

The % base is off the charts because of the sample types. But the run that worked was less even than the one that failed because the good run had 12% phix and the second failed had 23%

good %base


failed % base


The plot thickens, I was going to try running this library on a MiSeq across campus but that instrument has 2 fails on a RNAseq library running 2x250.

[GenoMax]

Can you comment on what is happening with phiX R2? Why is it not aligning (have you looked at the sequence, does it look random/with indels/N's)? What about phiX R1? You do have very low complexity samples.

[thermophile]

I haven't looked for the phiX seqs, but can do that. I was just looking at the % aligned, good run ~12% aligned both R1 and R2, bad run R1 23% R2 0%. When I look at the seqs for samples-there are ~20 N in the first ~40 bases. I can do pairwise blast for the R1 and R2 seqs, so the rest of the r2 is actual data.

I wasn't very clear about the structure of the amplicons, these are ~250bp so the first bases of r2 are the compliment of the last bases of R1. They are challenging libraries but this is our bread and butter library type.

[thermophile]

just started a 12pM phiX, 500 cycle run. I'll update with how read 2 of that one looks in 2 days

[thermophile]

well crap(ish). the phiX run looks fine. I'm waiting to hear back from my FAS with suggestions for moving forward with our major library type.

Anyone know how to find the lot info for reagent cartridges that you've thrown away? Is it in the run data somewhere?

[GenoMax]

Illumina should have that information, which your FAS should be able to get.

[microgirl123]

https://my.illumina.com/MyIllumina/Bulletin/ZOCgM9hxzkmvA-vfWNrXfA/how-to-determine-a-miseq-cartridge-lot-number-from

Illumina has a bulletin about determining lot number after the fact!

[thermophile]

sweet, thanks.

The second level team is looking at my last few runs. FAS thinks it's software related, that for some reason RTA is suddenly having issues with our low diversity libraries.

eta, the lot checker doesn't seem to work. It's searching (if I put in nonsense it gives an error) but never popping up the table with the info. Tried in IE and firefox just in case it was browser dependent

[Number6]

Did you upgrade the MiSeq software released around the end of November last year? We did, and all low complexity library sequencing looked terrible. We rolled back the software and all has been fine since.

[thermophile]

This is a new to me Miseq as of late Nov (replaced one that continually had focusing issues) We're running MCS 2.5.05. FAS was talking about reinstalling all the software but then decided that he didn't want to do that unless second level told him to. I haven't updated anything on the machine, but it's possible that someone else clicked an update button for something other than miseq controller.

Current plan is to run a mix amplicon and truseq genome run Monday.

[Number6]

The upgrade was to MCS v2.6; we rolled back to MCS v2.5

[GenoMax]

@Number6: Was the downgrade recommended by Illumina or something you thought of trying?

We upgraded multiple MiSeq in November and noticed that our machines started keeping the temp image files for some reason after the upgrade (this started ballooning the raw data folder size). Tech support just had us turn off the image saves as a workaround. They said that no one else had reported this.