Performance Evaluation of GPU-intensive Genome Analysis Workflows in HPC and Cloud

Presenter: Tomoya Tanjo, Bioinformatician and DDBJ Center, National Institute of Genetics, Japan; BioData Science Initiative, Joint Support-Center for Data Science Research, Japan
Co-authors: Yosuke Kawai, Shun Nogawa, Kentaro Yamamoto, Manabu Ishii, Tazro Ohta, Fumikazu Konishi, Tsuyoshi Hachiya, Osamu Ogasawara

Schedule:
Session 2 (Online Only) Thursday 9 May / Friday 10 May, APAC-Americas Friendly time
and
Session 5 (Amsterdam / Online hybrid) Thursday 16 May, APAC-EMEA Friendly time

NVIDIA Parabricks is a genome analysis toolkit that utilizes NVIDIA GPUs. To leverage Parabricks for polymorphism analysis, the National Center for Global Health and Medicine (NCGM) developed WGSPipeline in the Common Workflow Language (CWL). It can obtain harmonized results for typical chromosome regions. Thanks to the CWL, we can execute NCGM WGSPipeline in various computing environments with GPUs and container runtimes.

This presentation shows CWL features that make our workflows portable and shows experimental results with GPU nodes in the NIG supercomputer, which is the HPC environment in the National Institute of Genetics, and SAKURA internet, a domestic cloud environment.

Please leave your questions for the presenter below!

As an alternative to YouTube, this presentations is also available on ConfTube