Over the two days right before the annual ASHG (American Society of Human Genetics) meeting in Boston, Life Technologies held a two-day event where several hundred Ion Torrent customers and other interested parties attended.
Updates for three upcoming products (the Ion Chef automated template prep, the PII chip, and a product formerly called Avalanche but is now generically named ‘Isothermal Template Prep’) were shared which were expected. A new more highly accurate sequencing polymerase was announced.
For the Ion Chef, the first shipments to customers will be in December 2013, with a full launch in March 2014; for the PII chip, the first early access to customers will be in March 2014 with a full launch in May 2014. If you are not aware of what the Ion Chef and Ion Proton PII chip are, here’s something I wrote up in Sept 2012, and here is the Life Technologies webpage about the Ion Chef, along with the handy Product Specification Sheet.
The Ion Chef frankly has taken longer to get to this point, but the time has been invested to insure that the system works reliably and just as well as the prior systems the Chef replaces: the OneTouch 2 system for templating beads, the OneTouch ES for enrichment, and the best hands for manual loading of the chip. When you think about all the automation that it entails (not only replacing the above but also including barcoding of chips and an automated deck imaging system for visual sensing of deck components and reagents) the engineering involved is considerable. Undoubtedly earlier for the launch of Chef would have been better, but a later launch with a much more reliable and efficient system is better still.
The PII chip has gone through months of iterations, and in the last presentation two development scientists talked about what that involved. Apparently the three-series PGM chips (the 314, 316 and 318) as well as for the Proton PI chip there wasn’t the need for engineering the geometry of the well, but the PII chip (on the order of 0.3 µm if my memory is correct) there had to be a number of iterations to get a high percentage of wells to fill with the beads.
Also there was an acknowledgement of the precision and overall awesomeness of working with Dynal since late 2010 (Dynal is part of Life Technologies). They are able to make 450nm ISPs (Ion Sphere Particles) with a tolerance of +/- 20nm, which is remarkable. (I used to manage a magnetic sphere particle for QIAGEN for His-tag recombinant protein purification that was on the order of 60µm, and those were really difficult to manufacture at a consistent size.) And Dr. Mike McKenna said something interesting: “They are not really beads, more like microscopic jellyfish.”
The relative size of these wells is also remarkable – from 3µm for the PGM 314/316/318, to 1.3µm for the Proton PI, and now 0.6µm for the Proton PII. We now can’t call the PII wells ‘microwells’; rather we should be calling them nanowells instead.
Going into the engineering a bit further (back at UCLA if I didn’t go into microbiology as an undergraduate I would have gone into engineering instead), the PII wells were overlaid the PI sensor, so one out of four wells would generate signal, but that was the method by which the geometry of the wells could be iterated and the ISPs tested, and apparently at that scale there were many iterations of the well geometry.
Also a timeline was given about how the design of the chip wafer has been made final (some 30 mask layers, to give you an idea of the complexity involved) and sent out to manufacturing, a process that takes almost two months.
There was an interesting comparison of the expected data throughput of the Proton PII chip compared to maximum data throughput of other devices. A top-end commercial Cisco router can do 100 Gbps while the PII will do 120 Gbps, which represents 120 1 Gbps Ethernet cables (imagine that visual) or 36,000 simultaneous Netflix streams. The final slide showed the overall gigabase throughput progress, and in October reached 30 GB internally.
The product formerly known as Avalanche (apparently Life Technologies cannot use this as a name for the technology anymore) does not have a replacement name at the moment, but generically it is the Isothermal Template Prep, and will start early access before the end of the year. Lyophilized reagents will make the lab workflow that much easier (add water and resuspend), and it replaces making (and breaking) emusions, although enrichment is still required (so those OneTouch ES units will still be needed). A 2 hour process replaces a close to 5 hour one, so that will have a major impact on a laboratory’s daily scheduling. (It may not necessarily enable additional runs, but depends on that laboratory’s configuration and particular needs.)
Data was shown with large inserts – on the order of 500 bp for an E.coli library – with modal readlengths on the order of 350 to 420 bases and >99% accuracy. We were told that this process will port directly to both the PGM and Ion Proton.
Lastly, the announcement of the Hi-Q enzyme. Screening 10,000 polymerase mutants yielded a new polymerase that reduces insertion-deletion (“indel”) errors by some 90%, data was shown that summarized systematic errors reduced from 0.224% to 0.027%. An interesting slide was shown illustrating a 420bp run on a PGM, with all bases from 1 to 400 having a >Q30 quality. A dataset is now available from the Ion Community for those interested.
For those reading this far, I do appreciate all the positive comments and feedback just in a past few days. It’s humbling to meet people in real life who begin the conversation with an encouraging word about this blog. (And with this week’s ASHG meeting I’ll be sure to post some additional observations in the coming weeks.)