Costs of Running Production Genomics Workflows On-Premise as well as on Commercial Clouds

While the field of genome analysis is rapidly evolving, a number of tools have matured to become the de factostandard. The Garvan Institute has rapidly grown to become Australia’s leading genome-powered medical research institute, and this has meant that we have moved production genome analysis workflows out of the hands of individual researchers. Instead, these workflows are built by engineers to be run at scale on-premise, on commercial clouds (Amazon Web Services and Microsoft Azure) and at the National Computational Infrastructure (NCI). 
This project investigated the following questions: 
• Since Garvan built their 5000 core on-premise computational infrastructure using grants - what does it cost to run these workflows at Garvan, and how do these costs compare when running the identical workflows on AWS, Azure and the NCI. 
• What are the underlying costs of running these analyses that include, power, maintenance, staffing etc?