your experience with the Proton

[ajthomas]

We are considering buying either a Proton or a MiSeq. A little background: we're a small core facility at a university and we have a 454 only. We're running a lot of amplicons on the 454, but every other application has moved to other instruments. The advantage I see of the Proton is lower per-base cost and the ability to do transcriptomes (especially with the PII chip when that is available) and other larger projects. The advantage of the MiSeq is ease of use and longer reads (better for the amplicons we're currently doing, some of which are moving to the MiSeq in the near future).

So, I would like to know from you what has been your experience with the Proton? How reliable is it and how easy is it to run? Does yours get a lot of use and what kind of projects is it being used for? How does real-world performance compare to the published specs? Are you glad you bought it? Any other thoughts?

[MissDNA]

I really hope you guys have opted for MiSeq.

We have had the Ion Proton for a bit over a year now and so far nothing really good came out of it.

Answering your questions:
It is very easy to use, especially when you compare with 454.
We have used for whole genome, metagenome and transcriptome so far. Our best run reached only 9.4 Gb, mean read length 120 bp. We have had entire runs where not a single base reached Q30! Quality of the reads is really, really poor.
So yeah, real-world performance is not corresponding with published specs so far and we are not seeing any improvement.
We are not glad we got it, we wish we had gone with illumina!

[NextGenSeq]

We demoed a Proton and got very high quality sequence. The correlation of the RNA-Seq results with the same cDNA sequenced on a HiSeq was 0.998.

We bought it after seeing that.

[BrianJames]

We have had ours since July of last year. We have mostly run AmpliSeq Exomes, and whole transcriptomes. Our last 20 or so runs have all been >10 Gb (with many in the 13-14 range), Q20 bases 8-11 Gb, and average read lengths >150 bp. We did not start of with results like that, but that is what we are getting now.

Our problem runs can usually be traced back to sample quality, or human error (me). We are testing the Ion Chef for the PGM, and I can't wait until it is working with the Proton. It should eliminate a ton of work, and at least half of our sources of error (me).

I can't say what I would do with the current options. I would likely base it on the throughput of the facility, and cost of reagents.

Brian