I actually awoke today expecting an exciting press release, but I sure wasn't prepared for the big announcements from Ion and Illumina. Not that they were totally unexpected, but there's a huge difference between speculation and announced products (which, of course, are hugely different from ones you can actually buy!)
Ion lead off by announcing the Proton, an official moniker for what I and others had referred to as PGM 2. A larger chip sports 10-fold the number of sensors on the 318 (which was just officially released for sale). Ion is claiming this "Proton I" chip will enable sequencing two human exomes (coverage unspecified!) by this summer, when the system roles out. That claim probably assumes a 2X or so read length improvement by that time. At end-of-year they wish to launch a "Proton II" chip enabling an entire human genome sequence in a run. The instrument lists for $150K, but as with it's baby brother that figure is not an all-in price, and the requisite server will apparently sport a $75K pricetag. A bigger version of OneTouch will also be required; price unspecified. Presumably the whole package will be somewhere around $250K, which puts it under a GAIIx and well under a HiSeq, but with a run speed characteristic of Ion Torrent. The tagline is that a human genome will be obtainable in one day (start to finish) for $1K.
Illumina followed shortly afterwards by announcing the HiSeq 2500, which will enable generation of a human genome's worth of data in 24 hours, though no cost information was given and that time doesn't appear to incorporate library production. This will be available mid-year, along with upgrades to the MiSeq system. MiSeq will sport three times the data production, apparently partly via higher cluster density but perhaps other improvements are driving that as well. Furthermore, run times will be reduced. A 250x2 paired end kit will be released
A lot of bloggers have written nice analyses of these; both Proton and HiSeq 2500 have threads on SEQAnswers.com, Matthew Herper has a series of pieces (including a revision with a link to Ion's booth sporting a Proton at the Consumer Electronics Show!), James Hadfield has a nice piece as does Nick Loman and EdgeBio. I'm sure I'm omitting many others. Many of them had anticipated key pieces of these announcements.
There are a lot of things to see in this. Both announcements are going to further pressure on Complete Genomics, whose market cap is about its cash holding right now, indicating investor pessimism in their ability to succeed. Both companies seem to be betting that speed will be critical for clinical applications, a view I concur with. Both companies are offering upgrade options from their existing machines, $50K for Illumina and unspecified for Ion.
Proton will offer performance approaching a HiSeq, but with several distinct differences. Obviously, there is the chemistry difference which implies less expensive but with the homopolymer issue. Proton is a desktop machine, looking much like a laser printer. It isn't clear from the limited pictures whether Ion has designed them to stack. HiSeq 2500 is presumably of similar size to a HiSeq, which has a much larger footprint. The cost differential is sizable: $250K vs $800K (or so).
Ion's Proton I chip appears to just enlarge the 318 technology. However, Proton II will clearly require going to smaller sensors using smaller beads, for which limited proof-of-concept was demonstrated in their Nature paper. It remains to be seen how well this will work on a large scale: will well-to-well cross-talk become an issue, for example.
Nick's piece underscores that Proton breaks from the official line that Ion would allow continuous upgrades via new chips. Many were skeptical of this, and Proton proves it. Given the simplicity of the Ion hardware, the 3X higher price for the instrument is striking, and could well carry some significant profit margin. On the other hand, some additional cost would be expected to result from beefier fluidics and higher data collection requirements. The future of the PGM isn't clear, though I would expect it to stick around as an entry-level machine and one well suited for amplicon sequencing and some other applications. However, even if it sticks around it's possible it won't see a baby version of the Proton II geometry.
So summer 2012 will bring a new round of performance improvements for researchers. Whether these will be successful from a business point-of-view is another question. If academic funding remains flat (or worse), sales of the instruments or the probably more important reagents might stay flat as well. The threat of new entrants remains. NABSys recently made organizational changes intended to signal they are approaching commercialization, and perhaps Oxford Nanopore will finally get out of the gate. No matter what happens, those of us who thirst for fast, abundant & cheap DNA sequence will find much to be excited about.