Terrible overclustering -- 16S EMP protocol

[neko_33]

Hi SEQ-users,

We've been using the following protocol for 16S V4 region amplicon sequencing on MiSeq, which I think is very familiar to many of you:

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3400413/

For several years, most members in our lab have been constantly successful in that method.
Recently, however, cluster density is abnormally high -- about 1300～1500K/mm2(v2 kit) -- and sequencing quality very low. We tried over and over, reducing the library concentration from 15pM to 10pM, renewing sequencing primers, renewing phiX controls, but none of these worked.

In additon, libraries prepared with other methods (e.g. 2-step PCR, NEB UltraDNA Library Prep Kit, bacterial RNA-seq, etc.) are still sequenced fairly well. This probably means that, our trouble is not due to the MiSeq instrument or our poor sample handling.

Do any of you have such problems?
Or, does anyone knows the solution?

Thanks a lot in advance.

[nucacidhunter]

Most likely reason for overclustering in this case would be inaccurate library quantification and high clustering of low diversity library will result in low quality. If reducing loading from 15 pM to 10 pM reduced cluster number then you might review the quantification.

[pmiguel]

We recently had a set of libraries (not 16S, but targeted loci, so very similar) that read as 4nM on the bioanalyzer but 20nM via qPCR! These were so discrepant that we repeated both assays at least twice with similar results each time.
We used the qPCR numbers and ended up at ~700Kclusters/mm^2 on a v2 MiSeq cassette. A little low, but far preferable to overshooting the cluster density.
neko_33 what method are you using to quantitate your library concentration? Also, if you are using phiX as your standards for qPCR, don't do that! The phiX standards are notorious for varying in concentration from batch to batch.

--
Phillip

[neko_33]

quantification failure? -- maybe not

Thank you for your kind replies.

We usually quantify libraries with Qubit fluorometer, but at times with qPCR (KAPA Library Quantification Kits). Illumina Co. warns that Biolanalyzer quanitification is not reliable, according to my memory.

Empirically, not so much difference between Qubit and qPCR. For example, when I last did MiSeq run, the concentration calculated based on Qubit result was 1.9±0.1nM, while 2.2nM on qPCR.

As I wrote in the first message, I think our quantifications do not have thus fatal errors, but I will continue using both Qubit and qPCR, and compare these two results.

[thermophile]

Is this the same machine? Each machine clusters differently (I'd guess due to fluctuations in temperature control). FWIW, I cluster 16S runs at 6pM. When I go up to even 8pM I end up loosing so many clusters at PF that the output is roughly the same.

[neko_33]

Dear thermophile,

Thank you very much for your suggestion. The machine we use is always the same.
But I will check if the temperature is decently controled.

In addition, it is surprising that even 8pM library is too high!

[thermophile]

The EMP sequencing primers aren't the right TM (they're too low). I've had the opposite problem you've had in the past-regular libraries will sequence fine but I'll get very very few clusters on the 16s runs. This was on an older miseq and the temperature control module wasn't hitting and holding temps correctly. It was still within the error tolerance of illumina libraries but the emp libraries have a narrower tolerance.