Miles Benton from the Institute of Environmental Science and Research in New Zealand spoke at the Nanopore Community Meeting 2021 about “Affordable GPU compute makes nanopore sequencing even more disruptive and empowering.”Benton is a senior scientist, bioinformatician, and an advocate of open, reproducible science. They share their materials on GitHub. Benton spoke about the birth of his son 32 weeks premature. They had to do a lumbar puncture and testing, and Benton wondered about the accessibility of sequencing technology for diagnostics. Benton wanted to build a low-cost Nvidia GPU system for Nanopore basecalling. The Nvidia Jetson Xavier NX has a 6-core ARM, 384-core Volta GPU, 8 Gb RAM. The Xavier AGX is more powerful with 512-core Volta GPU and 32 GB of fast RAM. The AGX is capable of “keeping up” with high-accuracy base calling. Benton purchased a solar panel, hardware, and connecting the Mk1b. The community helped develop the “missing pieces,” and Benton ran a presentation and COVID basecalling at the same time! Benton went back to the the initial question: can you run basecalling on a $100 compute? Lim Fang-Shiang used a portable GPU system to learn about insect pathogens as a graduate student! Devin Drown and high school students printed a 3D case and got the Jetson Xavier NX unit to basecall. Benton shared that Lara Urban is using adaptive sampling for animal identification using mobile systems. Benton shared a couple of use cases of hardware to do point-of-care classification of tumors and AMR typing. Jurgen Hench has developed a prediction system for tumor samples based on methylation and CNV analysis that is near real-time and also using portable GPU systems. Matt Storey, Benton shared, has developed an application and interface to obtain results quickly without the need for high-performance compute. Benton shared that they developed a custom QC and Kraken-like pipeline for analysis that is no longer dependent on the cloud, which is of interest when working with human data. Benton concluded by going over specifications of several Nvidia GPU systems, highlighting their potential for use in the clinic and lab!
