NCM 2021: Affordable GPU compute makes nanopore sequencing even more disruptive and empowering
Miles Benton, from the Institute of Environmental and Science Research, New Zealand, began his presentation by discussing the inspiration for his work; he shared a heart-warming story about the birth of his premature son, who was suffering from suspected meningitis. Additional angst was driven by the lengthy turnaround times of the PCR test results. Being a scientist, Miles naturally questioned this, and referring to nanopore sequencing, pondered why, in the future, ‘we can’t have these amazing technologies at the bedside’. To that end, Miles recognised a potential avenue to explore — that of portable nanopore sequencing combined with affordable GPU compute to enable a robust, accessible real-time sequencing pipeline at low costs.
Underpinning Miles’ idea was the notion that GPUs accelerate basecalling, raising the question of pairing a cheap compute, like the NVIDIA Jetson series, with a MinION Mk1B. Speaking to the example use cases ‘which you take for granted being in the nanopore scene, you hear about a lot of these’, Miles ‘found that the user’s imagination really becomes the only limit’.
The fact that Oxford Nanopore products, such as the MinION Mk1C, integrate GPUs and software from the same family as the NVIDIA Jetson series, provided support for compatibility. Outlining some of the specs of these GPUs, Miles pointed out the Jetson NX and the more powerful AJX, are both capable of keeping up with at least a single MinION Mk1B in real time with fast basecalling. The Jetson AJX can just about cope with high accuracy basecalling in real time. The complete hardware, including the MinION Mk1B and Jetson, can be bought for less than $1,100, making it a highly accessible option.
Initially, Miles faced some obstacles in integrating the software, which he shared via Twitter. According to Miles, this really galvanised the community who chipped in to help piece together bits of the software to get it running. Highlighting the power and versatility of this system, Miles mentioned from the same GPU he was able to run his presentation whilst doing live COVID-19 sequencing at a conference, which you can imagine went down well. Confessing his admiration for the Jetson Nano, a $100 compute, Miles confirmed sequencing compatibility with the MinION — a really 'exciting' prospect.
Taking a step back, and looking at the broader picture, Miles next focused on the implementation of his work across the globe and the growing community that are heavily invested in further refining and improving methods geared towards democratising real-time nanopore sequencing using cheap compute. To highlight this, Miles referred to various examples of researchers using this platform for their projects. Lim Fang-Shiang, a Ph.D. student from Germany, reached out to Miles seeking his advice for combining a Jetson AGX unit to a MinION. Within just a few months, collaborators sought the help of Lim to identify a pathogen that was killing off many insects in their colony. Within just 24 hours, they had not only identified the pathogen but had fully assembled the genome at the site, an amazing feat that Miles modestly claimed, ‘it was really exciting to be just a tiny part of it’.
Next Miles mentioned his collaboration with Devin Drown, who has been working on a project with high school students, to leverage the MinION with the Jetson NX to perform adaptive sampling. Devin has since ordered more units to expand the outreach and get even more students involved, something which Miles is very passionate about; ‘getting people at an age where you can really empower them and excite them’. Although peripheral to the main focus of the talk, Miles felt compelled to confess ‘adaptive sampling is so cool’, and that ‘all these units are capable of adaptive sampling’, and Devon has demonstrated their success.
In a similar fashion to Lim, Stefan Prost reached out to Miles. He was conducting research into biodiversity in Columbian rainforests.
Introduced by Miles as another instrumental figure in driving the progress of the MinION paired with Jetson, Jurgen Hench, from the University of Basel has adapted the research introduced by Luna Djirackor at London Calling 2021 to classify brain tumours intraoperatively. The MinION and Jetson combination has obviated the need for high performance compute, and the Jetson AGX can run two MinIONs at once, quite comfortably — with live basecalling, and with methylation and CNV identification software running; ‘within an hour or two they can actually get a really nice classification on a tumour, which is just really exciting stuff’.
Going full circle, Miles recounts the moments of thinking ‘what could help identify what was in the cerebrospinal fluid of my son’ and recognises that now potential solutions are coming closer to reality. Addressing this, Miles discussed the collaboration with a local pathologist who was sold with idea, and after just one meeting bought a MinION and a Jetson AGX. Together they envisage a future where real-time diagnosis may be possible. Closing his talk, Miles suggests that this technology could be ‘game changing’.
Oxford Nanopore Technologies products are not intended for use for health assessment or to diagnose, treat, mitigate, cure, or prevent any disease or condition.