NEW YORK – Oxford Nanopore Technologies has experienced quality problems with some commercial batches of Flongle flow cells, its most affordable sequencing chips that were released earlier this year, and is trying to scale up production to meet rising demand from users. While dealing with these issues, the company recently announced that it will start offering sequencing products for regulated applications, such as clinical assays and food testing, next year.
Flongle users on three continents have reported that large proportions of their early flow cell shipments failed the quality control checks for the minimum number of pores available for sequencing — 60 out of a maximum 126 pores on the device, according to Oxford Nanopore's warranty.
David Eccles, a New Zealand-based bioinformatician who runs a consulting firm called GrinGene, said he tried out three flow cells in September, all of which failed to pass the 60-pore threshold. "The only [flow] cell I got any substantial yield for produced about 1,400 reads after loading, from one to four available pores," he said in an email. "Total yield was about 1.5 megabases."
He said he has heard similar stories from other users in person, on Twitter, and in the non-public Oxford Nanopore community forum. Three additional users, all based in Europe, told GenomeWeb of having similar Flongle flow cell quality problems, but did not want to go on the record about them.
Eccles and others said they were nonetheless able to get useful data out of Flongle flow cells that didn't meet the QC threshold and were drawn to the opportunity to provide feedback for the product design cycle. Also, at least two users said the main problem with Flongle wasn't performance, but the supply of flow cells, or lack thereof.
Oxford Nanopore declined to answer questions about Flongle on the record. The firm is scaling Flongle flow cell production to meet the initial — and continuing — demand but has not acknowledged the quality problems outside its user community.
The company may have found a solution that greatly improves the number of chips that pass QC checks, according to William Jeck, a pathologist at Duke University School of Medicine and an early-access user, who has been developing a gene fusion assay on Flongle with an eye towards the clinic. The first batches of Flongle flow cells were not delivered with the shipping buffer used with MinIon flow cells, he said. He estimated that about half of his first batch of approximately 10 Flongle flow cells failed QC. At some point, Oxford Nanopore then began shipping Flongle chips in the buffer. "We saw a dramatic increase in the quality of flow cells," Jeck said, with only two of his last 15 Flongle flow cells failing QC (overall, he has run approximately 35). Alexander Wittenberg of KeyGene, an Oxford Nanopore certified service provider, tweeted on Wednesday that his last four Flongle flow cells all had between 80 and 90 available pores.
Having flow cells that don't pass QC "isn't a huge deal, especially at the price point," which is around $90 per flow cell, Jeck said. "It's so damn fast, and with the high percentage of runs giving results, even meeting that 60-pore cutoff isn't that important. What's important is that you get sequencing data quickly."
At Oxford Nanopore's user meeting earlier this month in New York, CTO Clive Brown said demand for the devices continued to grow and the firm was scrambling to catch up to its order backlog. During his technology update presentation, a video of which the company made available on its website, he displayed a graph showing the gap between the "flow cells ordered" and "flow cells shipped" trendlines.
"You see we had a bit of a problem there," he said, eliciting laughter from the audience. "But in the past three weeks, we've trebled the production of Flongle flow cells," he said. "We think we'll catch up with demand by Christmas, easily." He also reiterated that the firm had had a problem with the GridIon due to a faulty third-party power supply, but said the issue had been fixed.
These issues, and Oxford Nanopore's responses, come as the firm is preparing to produce devices for regulated applications. At the same conference, CEO Gordon Sanghera announced the company will be launching a "Q" line of more regulated products in January 2020. As reported by Keith Robison, a researcher and nanopore customer who attended the user meeting, Q line will start with the GridIon and will later include the MinIon, along with several flow cells, including Flongle chips, and library preparation kits. The firm will begin by pursuing several ISO accreditations, including ISO 9001 and ISO 133485 certifications for quality management systems in medical device manufacturing.
Oxford Nanopore began taking orders for the Flongle, an adaptor that can be used with the firm's MinIon and GridIon sequencing devices and takes low-cost, single-use flow cells, in March. At its UK user meeting in May, Brown said the firm had achieved an internal record yield of 1.5 Gb, with one unnamed customer achieving 1.9 Gb, and that the median customer output was 524 Mb per run.
The company's initial plan was to sell the flow cells in packs of 48, along with one of the necessary adapters, for $5,250; however, by May the firm was offering smaller packs of 12, for $1,860. The adapter is priced individually at $750, and additional flow cells at $90. According to the graph in Brown's recent presentation, Oxford Nanopore began gradually increasing the number of shipped flow cells over the summer.
Since at least November, Oxford Nanopore has also been offering a promotion for Flongle flow cells with a lower QC threshold. Users willing to accept devices for which the firm only guarantees 30 available pores can receive two-for-one pricing. Oxford Nanopore did not comment on when and why it began offering the promotion, how many lower-quality flow cells it had available for sale, how it had determined that these should be guaranteed for only 30 available pores, or how long it planned to offer two classes of Flongle flow cells.
For the Flongle flow cells that do pass the quality control check, their performance, measured in sequence yield, is unclear. Because pores degrade with use, the yield is tied to the number of available pores.
A Flongle flow cell with the minimum of 60 available pores produced 1Gb of data during a cDNA sequencing run for Stella Loke, for example. Loke is an associate research fellow at Australia's Deakin University Genomics Centre who has used four Flongle flow cells so far. However, her three other flow cells did not pass the QC check, including one that had fewer than 10 available pores. "I got around 150Mb from a [whole genome sequencing] run with 49 pores," she said in an email, which had N50 read length of approximately 14kb, the same as she has achieved with MinIon flow cells.
Eccles said that some Flongle users were tripped up by an imprecise protocol on how to load the flow cells. Oxford Nanopore updated the protocol in mid-November, he said, and it seems to be helping more users find success; however, "there still remain unexpected drops in pores, particularly after loading libraries for sequencing," he said.
"The instructions have been very unclear," Loke said. On Twitter, she and other Flongle users have pieced together some recommendations. "The priming port must be free from storage buffer. [Oxford] Nanopore seems to be unaware that the flow cells have some leakage on shipping," she said in en email. The buffer can be cleared with a pipette and carefully chosen tip; similarly, the loading must be done very carefully and very slowly, she said.
Jeck said the storage buffer requires users to flush the flow cell before sequencing, which adds some hands-on time. "New users can have trouble with introducing bubbles into the flow cell during the flush," he said. "Any lab that runs more than a few of these is going to get that figured out very quickly, but it can generate some frustration in the first couple of runs if one is not careful."
Eccles said he has received free warranty replacements from Oxford Nanopore for the Flongle flow cells that failed quality checks (the firm's warranty requires users to report the failure within two days of the QC test.) He also said these kinds of issues are to be expected with a cutting-edge company like Oxford Nanopore. "They have mentioned previously that they'd prefer releasing products early so that the community can help guide development," said Eccles, adding that he prefers "seeing the early stages of product development and feeling like I have contributed to that."
"I especially like this when I can work out ways in which to get useful results out of the early products," he said. The results he did get out of the Flongle "were sufficient to answer some questions I had about the Oxford Nanopore rapid [sequencing library preparation] kit."
"Flongles that don't pass QC are still quite useful," Jeck said, "and they help me to characterize 'worst case' performance of the technology. For my diagnostic purposes, I find that even very poor QC flow cells can work very well, provided the libraries are prepared well." But he noted his purposes "are quite specific."
Loke suggested that "Flongle is good for very low-input work where you only have a tiny amount of library and the sample is precious. I would use it where the MinIon would have super low pore occupancy and it would be a waste of money." But for her own work, Oxford Nanopore's nuclease flush kits to extend MinIon flow cell yield have led her to set aside Flongle. "I would use MinIon to test a PromethIon library, then wash it, rather than use a Flongle," she said.
"I am a great believer in the potential that ONT offers," Loke said. "I just think that they release things a bit too early."