HPC Bioinformatics education: The University of Melbourne experience

The University of Melbourne is regularly assessed as one of the world's leading universities and much of this is a result of research strengths in the health sciences with several significant affiliated institutions. In the computational space, developments in the health sciences are heavily dependent on the broad and interdisciplinary discipline of bioinformatics. The use of bioinformatics applications, especially those with complex workflows or large datasets, is most effectively and efficiently conducted in a high performance computing (HPC environment). However, as well documentated in previous publications and presentations, many researchers are unfamiliar with HPC environments, often painfully evident in bioinformatics software. Basic education of these environments often becomes the responsibility of HPC centres themselves. 

At the University of Melbourne, there is a general-purpose HPC system, "Spartan", along with some smaller systems housed in affiliated institutions. Spartan's history starts with an innovative and cost-effective architecture and which now is a world's leading system with Top500 certification. Since Spartan's introduction a key feature of the enviroment has been an extensive user education programme, which is strongly correlated with very high levels of job submission. Specialist discipline-specific training workshops are also an on-going feature of the programme, including bioinformatics, with a training approach that is sensitive to the cultural backgrounds of the user community.

This presentation outlines the HPC bioinformatics teaching content and experience at the University of Melbourne. It is based on the recognition that HPC systems will continue to develop and incorporate new technologies that improve performance as data complexity and size increases. In fact, the evidence is that researchers increasingly require access to HPC systems and knowledge of how to operate in this growing environment. The University of Melbourne programme, covering operating system knowledge, application use, environment modules, job submission, web-based interactive environments, and parallel processing skills, uses a scaffolding approach starting from the very basics, to a distributed training programme with the "Spartan Champions" project. The successes of this programme is suitable for adoption by other institutions that also wish to educate researchers in HPC use, not just in bioinformatics, but other disciplinary areas.

ABOUT THE AUTHORS

Lev Lafayette is a Senior HPC DevOps engineer and the HPC Solutions Team Leader at the University of Melbourne, where he has worked for the past ten years. Prior to that, he worked at the Victorian Partnership for Advanced Computing in a similar role. Over the past twelve years, he has taught thousands of researchers across more than twenty Australian Universities and government agencies. An incorrigible collector of degrees and with a wide range of interests, Lev has just completed his fourth master's (in Climate Change Science and Policy) and has eight degrees to his name. He has just started a doctorate at Euclid University.

Daniel Tosello is a HPC DevOps engineer at the University of Melbourne, where he has worked for the past ten years. Prior to that he worked as a software engineer at La Trobe University and for ICT client support and also as an IT support officer at the AARLIN Consortium. Daniel has recently presented at eResearchAustralasia with representatives of Victorian health organisations on tailoring HPC for medical research. He also has interests in 3D printing and CNC pipelines.

-----

For more information about the eResearch NZ / eRangahau Aotearoa conference, visit:
https://eresearchnz.co.nz/