Education, HPC, and the Cloud.

Our three part series on utilising the cloud for education in supercomputing kicks off with a look at the CompBioMed Center of Excellence.

Part One: Setting Project Foundations

For over five years the Alces Flight crew has been studying where cloud works best within a HPC service. In this series we’re exploring cloud HPC in education through our work with the CompBioMed Center of Excellence and their cloud HPC training cluster, nUCLeus.

Adding Cloud to a Consortia.

The CompBioMed Center of Excellence focuses on turning the science fiction of personalised medicine into the science fact of everyday clinical use. With consortia membership ranging from purely academic, to start-ups, SMEs and corporates the overarching goal is one of collaboration for a better future in medicine through the use of computational biology.

Thanks to a strong cooperative foundation CompBioMed has access to some of the word’s most powerful HPC clusters — conducting research and hosting events to educate and empower their membership. When looking to add cloud to the consortia the training and education team focused on growing the mission of empowerment by expanding access to learning. They chose to consolidate their knowledge on core foundations in HPC — something that any student or researcher anywhere could undertake as a starting point in their journey into supercomputing. By using a complimentary approach to cloud CompBioMed is expanding their resource footprint without disrupting ongoing research. In fact, it is hopeful this method will lead to a wider interest in their work and encourage more research in the field.

Demystifying HPC for Biomedical Research.

Thanks to recent advances in the capabilities and availability of HPC and Artificial Intelligence (AI), concepts in personalised medicine that once sat on a desktop are now being ported to and used on HPC clusters. A key research topic for CompBioMed is how the processing of these workflows can be automated and optimised as students move from running them on smaller workstation-class machines to much larger, more capable HPC clusters — something that is currently considered a ‘dark art’ from the perspective of a researcher with little familiarity with HPC.

As supercomputing develops the potential to be as much a tool of the trade as the pipettes and test tubes that biomedical researchers use day-to-day, it is important that researchers are appropriately introduced to its use and understand how to optimise application parameters to best meet the needs of their scientific question.

Broadening the HPC user base — the QIIME2 Project.

The education and training team at CompBioMed selected the QIIME2 application as their first foundation course within their broader learning objectives. QIIME2 was chosen due to its low-intensive computing nature, its well-documented history of being ported from desktop to HPC, and it’s broad popularity across research and clinical work.

For this initial launch the academic institutions UCL and the University of Sheffield came together to collaborate on the course which would be presented as a a Student Selected Component (SSC) within the undergraduate medical degree at the University of Sheffield. Building off previous coursework, they created a class designed for medical students keen on adding computational knowledge to their studies. This six-week, self-selected course would be not only a test of cloud HPC capabilities, but one of working together across multiple sites to ensure no student was left out of learning. The course quickly reached full enrolment — and the team set about placing three core goals into action.

CompBioMed’s core goals.

CompBioMed had three key requirements defined early in their cloud and education design that were necessary in making the nUCLeus cloud HPC successful:

  1. Keep it complimentary — their cloud development is designed to compliment their on-premises access by focusing on engagement with new users and new usage requirements which have or have potential to progress to production systems and platforms.

Part two in the series will cover how nUCLeus was designed, architected and scaled to meet these core goals. You can read it right now, or, if you would like to talk with us more about your cloud HPC goals why not get in touch?

Software for research computing