SAN FRANCISCO (GenomeWeb) – Early customers of Oxford Nanopore Technologies' PromethIon instrument are planning experiments ranging from phasing clinical human genomes and understanding the role of structural variants and epigenetics in neurological diseases to metagenomic sequencing. Many of these users indicated that although initial runs were of lower throughput and quality than anticipated, performance has improved with each run.
The current version of PromethIon in most labs' hands is the alpha device, which can run 24 flow cells in parallel. Most researchers are still in the very early stages of testing the device and have only run fewer than a handful of flow cells in parallel. Instead of maximizing the number of flow cells, they have primarily been focused on increasing their flow cell yields — regularly setting and breaking new records in a mini-competition taking place on Twitter — and have reported yields between 50 gigabases and nearly 100 gigabases per flow cell.
This week, Oxford Nanopore began shipping its first beta systems, which will initially run up to 24 flow cells simultaneously, each capable of generating 150 gigabases of data, and will eventually enable users to run 48 flow cells in parallel, although the company has not specified a timeline for that. Also, last month the company described improvements it made to its flow cells to boost performance.
Researchers from the Wellcome Centre of Human Genetics in the UK, the Australian Genome Research Facility, and Belgium's VIB discussed their experiences with the system thus far and how they plan to use it in the future.
Ken McGrath, manager at AGRF's Brisbane location, said that following a few snafus with setting the instrument up, perhaps due in part to not reading the instructions, the system has been continuously improving, particularly concerning throughput. The lab is now generating around 70 gigabases of data per flow cell, he said.
Library prep has been "straightforward and simple," McGrath said. But dealing with the data coming off the instrument has been complex, as it currently requires a lot of IT support to monitor reads, he said. However, he anticipated this would improve and attributed those issues in part to the "phenomenal amount of data" being generated, even in the first few minutes. "There are challenges in moving that data around," he said.
McGrath said the data quality is comparable to the latest sequencing chemistry of the MinIon. At the Advances in Genome Biology and Technology meeting in Orlando, Florida last month, he compared data from the PromethIon with data from Illumina's HiSeq for sequencing a 10-organism metagenomic reference sample. The two systems generated different types of reads, and while the PromethIon's longer reads were better able to identify organisms down to the species level, the sheer volume of Illumina reads were better at picking out rare organisms.
He said that as a sequencing core facility, the lab would use the instrument for "anything under the sun," depending on what customers wanted. Currently, he said, the lab is testing out various applications to identify its sweet spots, including sequencing human, bacteria, and animal genomes. "As data yields increase, I think we'll approach a point where de novo genome assembly is possible even for large genomes," he said.
Researchers at the Oxford Genomics Centre, the genomics service facility of the Wellcome Centre for Human Genetics, also had a rough start with the PromethIon. David Buck, head of high-throughput genomics, said his group was the first to receive a PromethIon and the initial runs were "pretty dire," with the flow cells generating "poor amounts of data." But, "the last four flow cells we've run were phenomenal," Buck said, each yielding between 54 gigabases and 76 gigabases of data. Like McGrath, he said the system's software and hardware have been challenging.
The lab has sequenced a clinical human genome sample on the PromethIon, he said, and that data is in the process of being analyzed. The group has been focused on increasing yield rather than generating ultra-long read lengths. "When you go for the ultra-long reads, the output goes down significantly," he said.
He said the technology is "headed in the right direction," but its "Achilles heel is that it's not very accurate." Nonetheless, Buck said he sees potential in the technology for applications like phasing genomes and identifying structural variants. It could also be combined with Illumina sequencing, which would improve the accuracy and enable SNV calling. In addition, he said, it could be a great tool for transcriptomics, which is primarily a counting application.
At Belgium's VIB, a life sciences research institute that works with five universities in the country, researchers have already begun using the instrument in their research projects. Halina Novak, manager of technology and innovation at VIB, heads a laboratory charged with evaluating new technologies, including the PromethIon. Wai Long Tam, a life sciences technology specialist, has been spearheading the evaluation, and the lab has begun collaborating with researchers who want to use the technology in cancer, neurology, and metagenomics research projects.
One advantage of the device, Tam said, is that it "lets us scale to our needs." For projects where read lengths of 10 to 20 kilobases are necessary, the DNA can be sheared to that length and the majority of reads will be that size, "so it's very flexible," he said. Similar to other users, Tam said that the flow cell yield has been continually increasing. Initially, the yield was 10 to 20 gigabases, but with the last few flow cells, yield has been well over 50 gigabases. "Depending on the library, 80 to 90 gigabases is no longer a surprise," he said.
The lab has also compared the PromethIon with the MinIon and found that the data is comparable. "It's an upscale in the output rather than differences in accuracy," he said.
Mojca Strazisar, a scientist at the VIB Genomics Core and the University of Antwerp Center for Molecular Neurology, has been working with Tam's group on the PromethIon to study structural variants and epigenetics in neurological disorders. She's hopeful that the most recent run, which is still being analyzed, will surpass the 100-gigabase mark, which has become the end goal of an ongoing competition between PromethIon users playing out on Twitter.
She said it was too early to comment on the data itself. "We can see that we're detecting structural variants, but it's too early to say whether they are true positives or false positives."
But, she said, the technology is attractive for both its long reads, which will enable structural variant detection, as well as its ability to see epigenetic marks in native DNA. She said that if the instrument performs the way it's supposed to, the group plans to use it to start screening cohorts of patients with neurological diseases such as Alzheimer's and dementia to "start trying to unravel the genetic and epigenetic factors" that may lead to the diseases.
Two key improvements, she said, would be to continue to increase the output of the instrument and to improve the stability of the pores, so that they could sequence for a longer amount of time.
A number of researchers testing the PromethIon think it will likely compete with both the Pacific Biosciences Sequel system and 10x Genomics' Chromium instrument, although they said it was still too early to tell. With the PromethIon, Oxford Nanopore aims to drive the cost of sequencing a human genome below $1,000, and CTO Clive Brown has said that reagent and consumables costs for sequencing a human genome on one 90-gigabase flow cell would be just under $800.
Oxford Genomics' Buck said that although he does not currently have a PacBio's technology, his lab runs the Chromium and plans to run the same samples on the various platforms to "see what the optimal solution is" for clinical human genome sequencing.
Tam said that the VIB center has all three systems, as well as Illumina sequencing instruments. "We're quite happy with the different sequencing devices we have," he said. Novak added that she anticipates the lab will ultimately use different technologies for different applications, but said the group is still figuring out each system's sweet spot. "It will depend on the organism and the research question," she said.