Ion Torrent Proton Announced: The Chip Is (Not) The Machine
10 Jan 2012Big announcement today from Life Technologies who have announced the Ion Torrent Proton(tm). This, to all intents and purposes is the PGM-2, the second iteration of their pH-meter-on-a-chip sequencing machine. Indeed it is curious they have dropped the Personal Genome Machine moniker, as this product is the first to explicitly target the $1000 (human) genome market in benchtop format. So what do we know so far?
* Mid-2012 launch for early access customers, Q3 for general release
* $149,000 list price (compared with $50,000 for the PGM)
* Two chips announced
* Ion Proton 1 available at launch - 165m sensors (cf ~12m for the 318 chip and ~1m for the 314 chip)
* Ion Proton 2 in early 2013 - 660m sensors
* Update: 14:28 GMT - According to GenomeWeb, Ion Proton will also require purchase of a $75,000 server
[caption id="" align="alignright" width="300" caption="The iPod dock is gone, as have the milking tubes"]
[/caption]
Eric Olivares over at Seqanswers.com already did the critical calculations. Assuming a 250bp read length you could expect between 3 and 6 exomes at 100x coverage with the Proton 1 chip, and between 1-1.5x whole human genomes at 30x with the Proton 2.
How will this be acheived? Looking at the picture the chip size is visibly larger, but the increase in area (assuming the picture shows normal sized hands!) is not sufficient to account for the increased number of sensors, implying the wells are closer together and/or smaller than existing Ion chips.
[caption id="attachment_952" align="alignleft" width="300" caption="Proton chips are bigger than the previous generation"]
[/caption]
So, what to think about this announcement?
Well, the specs are certainly impressive and this is the first concrete evidence demonstrating continued log-scale improvements using the Ion Torrent technology after the 318 chip, something the Life Tech roadmap has deliberately not spelt out until now.
However, I have to admit to being a bit disappointed that this is an entirely new machine coming in at three times the cost of the original machine. It turns out unlike what we have been told countless times at presentations, the chip is not actually the machine. Whilst no surprise to anyone that has looked at the technology carefully, this may be a disappointment to those who just received or ordered their Ion Torrent PGM and have bought into the marketing wholesale. Are customers now looking at taking delivery of a machine that has hit its performance ceiling as early as mid-2012? We don't know how this announcement affects the PGM1 roadmap, it would be nice to get reassurance on this point. Update 14:31 GMT - GenomeWeb report an Ion Torrent marketing director as saying "The electronics on the PGM were built to handle chips of 1 million to 10 million sensors", which seems like confirmation that the 318 chip hits their density ceiling. However read lengths can still increase to generate additional throughput.
This announcement also has resonance for Ion Torrent's constant comparison to the computer and semiconductor industry. Was it really a wise move to announce a PGM-killer so early into this machines life? I am thinking specifically of the Osborne effect. This is the act of announcing an improved model early, in this case Osborne with their early portable computers, "to reassure current customers that there is improvement or lower cost coming, to increase the interest of the media and investors in the company's future prospects, and to intimidate or confuse competitors". The announcement can actually have the an unintended effect that customers see something better on the horizon and defer their buying decision.
If I was about to buy a PGM, would I now hold off for the Proton? Life Tech will be keen to target the PGM at cheaper "small genome" applications and the Proton at human genome scale, and this will be fine for many users, perhaps those wanting to replace amplicon sequencing done on capillary sequencers with high-throughput instruments. Me personally, I'd always want to buy the most fully-featured instrument available, akin to choosing a new laptop (which you know will be out of date before it is switched on for the first time). In bacteriology we are moving from 16S amplicon studies and whole-genome studies of single isolates to sequencing of entire communities as metagenomes. Such studies benefit from as much throughput as possible.
It will be fascinating to see how this announcement plays out during the year, particularly so close to AGBT where there is reasonable expectation some new "killer" sequencing platforms will be announced.
Have you just bought a PGM? What do you think? Leave a comment below. Also please head over to Seqanswers.com to read more discussion.
Happy New Year!