Ion Torrent Proton PII-PIII Chips?

[scubachris]

Hello,

apologies about the simplicity of the question but I'm a bit out of the loop and I couldn't find any relevant info online. Back in 2013 I used to work at a lab that had an Ion Torrent Proton, and I remember reading about the upcoming PII and PIII chips that supposedly had much higher throughput than the PI. Does anybody know what happened to them? Has the Proton been completely replaced by other instruments?
Thanks in advance.

[Bukowski]

I'm not sure the PII chip ever saw the light of day. The platform was effectively discontinued with the launch of the S5:

https://www.thermofisher.com/uk/en/home/life-science/sequencing/next-generation-sequencing/ion-torrent-next-generation-sequencing-workflow/ion-torrent-next-generation-sequencing-run-sequence/ion-s5-ngs-targeted-sequencing.html

And depending on your use case they may be better platforms on the market

[r.rosati]

I work in a lab with an Ion Proton.
Despite chats that "some laboratories" were already testing it, the PII chip was never released.
What was released, for the new S5 platform, is the Ion 550 chip; which should generate 100-130 million reads, as opposed to the 60-80M "official" reads for the PI (although we constantly get 80-90M reads on PI chips). This is not a chip that allows human genome testing as "promised"; it's just a chip that somehow comfortably fits 3 exomes instead of PI's 2 exomes per chip.
As of the platform:
Within the scientific community, my feeling is that it has definitely fallen out of grace in research settings. And whatever grace there ever was, I must say. This is mostly because the platform shows non-random errors, located at specific sites (including but not limited to homopolymer sites) which makes those sites basically un-testable with the technology. Thermo is still trying to reduce this issue. The new Hi-Q kits perform quite better, but the problem persists. And lately, Thermo made some adjustments to their software which look more of a "hiding errors under a rug" instead of actually trying to improve the technology; which was a bit worrying to witness. They recently introduced "effective regions" in their targets file, aka the variant caller plugin will simply cut away and ignore any portion of reads that fall in introns just a bit farther than the splice donor and acceptor sites, in hope of ignoring lots of homopolymer regions within introns. Thermo also recently increased - in my opinion, a bit silently - the minimum number of reads needed to call a variant in their standard VariantCaller parameters, from 10 to 35-40 minimum reads. I'm not a big fan of this, as I prefer to filter away a false positive than to get a false negative. I only noticed it 6 months after the change; so now we have to apply one custom parameter instead of relying on the (in theory, optimized) factory-suggested ones.
In diagnostic settings, Thermo has a few aces up their sleeve with Ampliseq, but Thermo & Illumina made a deal so that Ampliseq technology can now be used on Illumina. This is good for Thermo, but unfortunately the prices for kits have spiked a bit, so much that for a small lab, having an S5 or Ion Proton can be pointless unless you can wrestle some discounts; it's just cheaper and better to outsource and send your samples to be sequenced with an HiSeq.