NEW YORK (GenomeWeb) – Oxford Nanopore Technologies today provided an update on its business and products, including the pending release of the Flongle flow cell adapter, a compute module for the MinIon called MinIT, improved PromethIon flow cells, and efforts to improve the consensus accuracy for all of its platforms.
Separately, the company said that the US International Trade Commission has determined its products do not infringe patents held by Pacific Biosciences.
During a webcast, Chief Technology Officer Clive Brown said that the company has more than 6,000 MinIon customer accounts now – double the number of MinIons that were out in the field in late 2016 – and that it typically sells several hundred MinIons per month. Also, the firm has received more than 100 GridIon orders since the launch of that platform last year, most of which have shipped. Unlike the MinIon, the GridIon, which can run up to five flow cells in parallel, can be used for fee-for-service work.
China is becoming an increasingly important market for Oxford Nanopore. Last year, the company named a Chinese distributor for the GridIon and said it was preparing to open a new office in Shanghai. It has already sold several GridIons in China and is now selling the MinIon directly through its online portal. Chinese customers can expect delivery of MinIons within 72 hours of their order, and the first instruments are scheduled to arrive after the upcoming Chinese New Year holiday, Brown said. Some Chinese customers are offering sequencing services on the GridIon, and several nanopore sequencing publications have come out of China already. "We are hoping and anticipating that MinIon will be as popular in China as GridIon is," Brown said.
The PromethIon, Oxford Nanopore's high-throughput sequencing platform that is designed to run up to 48 flow cells in parallel, has been available since last year through an early-access program and is now fully commercially available.
One factor that has delayed its adoption has been the quality of the PromethIon flow cells, Brown said, but the company has now overcome this and is shipping high-quality flow cells to customers. Current early-access customers have alpha or alpha/beta systems, which the company will replace with beta systems starting this month.
Internally, the company has been getting well over 100 gigabases and up to 125 gigabases of data per PromethIon flow cell, he said, and yields of 80 to 100 gigabases on a regular basis. It recently started using an "active unblock" technology that clears clogged nanopores by reversing the voltage after software detects a blockage. That feature, which Brown said can be rolled out to customers quickly, has been nearly doubling the yield in the company's latest runs, and it expects to obtain 200 gigabases on a single flow cell within the next few weeks. The best customer PromethIon runs so far have generated 50 gigabases per flow cell but ongoing runs might top that, he said.
Based on the company's estimates, users will be able to sequence a human genome at 30X coverage for $800 on the PromethIon in the first half of this year, a price that will requires them to run the machine at high throughput to obtain volume discounts.
Brown reiterated that the PromethIon, not the MinIon, has been designed for human genome sequencing, though a recent paper in Nature — published more than a year ago as a preprint — used the MinIon to sequence a human genome. Though it might be possible to bring the price of a MinIon flow cell down far enough to enable a $1,000 genome, that would be a "very difficult proposition," he said.
On its website, the company now offers the PromethIon system for a bundle price of $135,000 that includes 48 PromethIon flow cells ($69,600), a starter pack of 12 PromethIon flow cells and kits ($28,575), compute service ($15,000 per year), reagent kits ($8,456), 12 MinIon flow cells ($8,100), and shipping and installation ($5,269). It does not charge customers for the actual instrument.
The Flongle — short for "flow cell dongle" and first announced in 2016 — is an adapter for both the MinIon and GridIon that contains sensing electronics and can be loaded with small, inexpensive, one-time-use flow cells. The goal is to open nanopore sequencers to additional types of users. "We think that there are a lot of people who will do more things if they are quicker, easier, and cheaper," Brown said. The device will be particularly suitable for clinical use, he said, and the company now has a commercial team that is pursuing clinical applications for its products.
Flongle will enable both DNA and RNA sequencing and is designed for rapid or small-scale applications, such as targeted sequencing, sequencing of small genomes, sample quality checking, clinical tests, environmental monitoring, or agricultural applications, according to the company.
The introductory price per Flongle flow cell will be $90 to $100, Brown said, depending on order volume, but there might be different pricing schemes in the future. Each Flongle flow cell contains 128 nanopore channels, a quarter of the MinIon's 512 channels, and the anticipated data output will be 1 gigabase to 1.5 gigabases in 10 to 16 hours, which could increase to 3 gigabases over time, Brown said. The sample loading volume will be 30 microliters, and samples can be loaded onto the Flongle flow cell without a pipette. The goal is to integrate sample and library prep into Flongle flow cells in the future.
Flongle will be available to early-access customers in the second quarter, around the time of the company's user meeting in late May, and a full commercial launch is planned for the third quarter.
The company has also been developing a compute module for the MinIon, called MinIT, that it also plans to make available starting in May. MinIT, which will cost $2,000, replaces the need for a dedicated laptop for the MinIon, and the company hopes it will further drive the adoption of the MinIon. "We noticed that GridIon is easier to install, configure, and run than MinIon," Brown said. "For a lot of users, the laptop is a significant barrier to getting the MinIon."
MinIT is a small box, weighing about a pound, that comes pre-installed with software, including the MinKnow sequencing control software and a Linux operating system. It also performs basecalling, generating Fastq files that are stored on a portable drive and can later be transferred to a computer. The system is in late development and is "very close to keep up with full-on MinIon runs," he said.
In addition, company researchers are working on "read-until" workflows that will allow users to keep a MinIon or GridIon going until they have collected sufficient data for a particular application. These workflows, which will allow users, for example, to select only very long reads, will be able to run on the MinIT.
Brown also said that the company has been improving existing kits and flow cells. In March, it will come out with a new kit that will provide longer average reads and higher throughput. That kit will use a fragmentation-free protocol, incorporate DNA repair, and enrich for longer DNA fragments. "We're actively developing long and ultra-long [read] methods whilst maintaining very high throughput," he said.
In addition, the company maintains a "very active research program" to improve sensitivity, he said, so users can start with nanogram amounts of DNA. One of the problems has been that the tethers on the DNA, which make it stick to the flow cell membrane near the nanopores, also make the DNA stick to other surfaces. This, he explained, can lead to the loss of up to 90 percent of the DNA before it gets "anywhere near the pore." His team have been working on a new technology "where the tether is within the flow cell," he said, so no DNA gets lost on the way. "Apart from boosting sensitivity, that new approach will dramatically drag up the tail of bad runs" seen by some users, he said.
The company has also been working on improving its consensus accuracy. Most sequencing errors happen at the beginning or end of homopolymers, Brown said, while some stem from DNA methylation or DNA damage, and the firm has identified three strategies for avoiding these. One is to improve basecalling algorithms and how they are trained, as well as to optimize run conditions.
A second approach is to use different nanopore proteins. Brown said the company has identified one protein, called R8, which has a long barrel that spans a large number of bases. Homopolymers that completely fit into this pore generate a "non-flat" signal, he said, that can be read. Another protein, R10, which was published by Oxford Nanopore Cofounder Hagan Bailey several years ago, has two pore constrictions and gives a "very clean signal," Brown said. The company would most likely maintain the existing pores but would offer flow cells with either the R8 or the R10 in addition, he added.
Thirdly, a fifth type of base could be incorporated during sample preparation, which would break up homopolymers, so they could be called and translated back into a 4-base sequence. "All three approaches are under active development," Brown said. "We're pretty confident that one or all of these approaches will close off the homopolymer consensus issue in the first half of this year."
Direct RNA sequencing, which Oxford Nanopore launched last year, has turned out to be a very popular application, Brown said. Customers have reached up to 5 gigabases per flow cell with RNA so far, which is less than with DNA because a different, slower motor protein is used to control the movement of RNA through the pore. Brown said the company is "working actively to find a new motor protein that will run faster." Another area of development is RNA basecalling, which is more complex than DNA basecalling because RNA bases are often modified.
Several other products in development — including the SmidgIon sequencer, Zumbador and VolTrax sample prep devices, the company's DNA synthesis platform, and a new single-cell sequencing platform the company is developing will be discussed at Oxford Nanopore's "London Calling" user meeting in May, Brown said.
The company is maintaining its longstanding policy of not selling its platforms to competitors, he said, in order to prevent other companies from reverse-engineering its technology.