NEW YORK (GenomeWeb) – Oxford Nanopore Technologies today launched the GridIon X5, a new desktop nanopore sequencer with throughput between that of the existing MinIon and the high-throughput PromethIon platforms.
During a live webcast today, Clive Brown, the company's chief technology officer, talked about the new sequencer, which uses the same types of flow cells as the existing systems, and provided updates on a range of other products, including the PromethIon, base calling algorithms, and 1D2 reads.
The GridIon X5 will be able to run up to five flow cells at a time, enabling it to generate up to 100 gigabases of sequence data per 48-hour run with current chemistry and software. It comes with a built-in computer for real-time base calling. Overall, the system, which weighs 10 kg, appears to be quite different from the original, stackable GridIon that the company first talked about in 2012 but then shelved as it focused on the launch of the MinIon.
The platform is available under two pricing models: using the capital-loaded pricing model, customers can pay $125,000 upfront, which includes one year of support, and $299 per flow cell, translating to $15 to $30 per gigabase, depending on the output per flow cell. Under the consumables-loaded pricing model, they can pay $15,000 per year for support and $475 per flow cell, translating to $23.75 to $47 per gigabase.
Customers are also allowed to offer sequencing services on the GridIon X5, which they are not permitted to do on the MinIon under the current license. For an annual fee, Oxford Nanopore offers a certification program for nanopore service providers for the GridIon X5 or PromethIon, which includes training and validation.
Oxford Nanopore is taking orders for the GridIon starting next week and plans to ship the first instruments by May, Brown said.
In the meantime, early customers of the PromethIon, which the company started shipping last year, can expect to receive their first flow cells in early April. On April 3, Oxford Nanopore plans to start shipping flow cells to at least 12 and maybe more customers who have received and configured their platform. Typical output right now is 10 gigabases per flow cell in a six-hour run, he said, and customers will be able to obtain 50 gigabases per flow cell in longer runs. At the moment, Oxford Nanopore ships between one and two PromethIon systems per week, he added, and should have its back orders cleared by the third quarter.
Brown also provided a timeline for PromethIon hardware development, which includes a design change. The systems currently being shipped are considered alpha systems and can run up to 24 flow cells in parallel, he said, half the number of flow cells the system can accommodate. Data from these will be written to an external disk for base calling.
In the third quarter, the company plans to release alpha/beta systems that will have a more powerful compute module in a separate, larger box and still run up to 24 flow cells in parallel. The new compute model will be compatible with real-time base calling at sequencing speeds up to 1,000 bases per second, so it will be able to keep up with future improvements in sequencing chemistry. The computer can also be used for local bioinformatics tasks when the sequencer is idle.
In the fourth quarter, the company will then enable all 48 flow cells to run in parallel, and in 2018, it plans to launch a fully integrated PromethIon system, called Mk 1. Customers who already purchased the system will obtain all these upgrades.
At full capacity, the PromethIon has a theoretical maximum data output of 11 terabases per 48-hour run, or 233 gigabases per flow cell, Brown said, so "it's not going to go out of date" even if it takes the company a while to reach those performance specs. For the early access program, output will be around 50 gigabases per flow cell, or 2.2 terabases in total, he said. In the future, run times can be extended to 4 days, he said, which will increase output.
The company has also made progress in improving the accuracy of nanopore reads on several fronts. For example, it has been developing so-called 1D2 sequencing, where it reads both strands of a double-stranded template without having to connect the two strands with a DNA hairpin, which it has been doing for so-called 2D sequencing. Under the new scheme, the second strand is held on the membrane near the pore while the first strand is being sequenced. Once the first strand exits and the pore opens, the second strand enters and gets sequenced as well.
Brown said 1D2 read sequencing will be released to customers that are licensed as developers immediately, and more broadly around the time of its UK user meeting in May. He said the accuracy of 1D2 reads is significantly better than that of the old 2D reads, which will be discontinued in early May.
In addition, the company has been working on improving base calling. A new base caller it developed, called Scrappie, can now call homopolymers more accurately than previous callers, Brown said. Oxford Nanopore now offers three types of base callers: Albacore, its fully supported production base caller; Nanonet, a research base caller that is available as an open source tool on Github and has limited company support; and Scrappie, another research base caller that is currently only available to developers and also has limited support. Base calling in the cloud will be discontinued on March 21.
In addition, the company has been working on base call acceleration by mapping its algorithms onto different processors and has been collaborating with Intel on that.
Finally, nanopore read length keeps increasing, Brown noted, largely due to improved DNA extraction methods that leave the molecules intact. One customer, Nick Loman and colleagues, recently reported obtaining a 950-kilobase read, he said, and sequencing entire human chromosomes will likely be possible in the near future.
However, customers typically get between 3 and 11 gigabases of data from a MinIon flow cell, Brown said, whereas internally, Oxford Nanopore obtains about 20 gigabases. Some of this customer variation is due to DNA extraction and sample type. In addition, he said, the DNA-helicase complex sometimes gets wedged in the pore, thus blocking it. To get rid of this problem, the company has worked out a way to reverse the potential for a short time, a deblocking method that will be implemented in the next version of the MinKnow software.
At least 4,000 MinIon sequencers are at customer sites all over the world, he said, including in China, India, and Japan. The company has no plans to revise the current Mk 1 MinIon model for the foreseeable future.