→ Poster doi:10.7490/f1000research.1112739.1
Eric Enns1, Philip Mabon1, Daniel Bouchard2, Mark Iskander2, Gary Van Domselaar1,2
1 National Microbiology Laboratory, Winnipeg, Canada
2 University of Manitoba, Winnipeg, CanadaAbstract
Galaxy has been in use at Canada’s federal public health laboratory, the National Microbiology Laboratory, since 2010. Prior to our incorporation of Galaxy, all bioinformatics tools were run manually on our cluster by bioinformaticians and a select few biologists. As we have transitioned to Galaxy as our primary computing environment, more biologists have been empowered to run their own analysis, which has increased the load on our cluster. To prevent job failures, Galaxy was configured to request a static amount of resources per tool which was suboptimal.
Galaxy allows tools to use a dynamic destination, which permits resource optimization. A survey of available dynamic destinations revealed that these are specific to a tool. Rather than develop a specific dynamic destination for every tool we have installed, our objective was to develop a universal dynamic destination solution that would work with every tool.
To this end, we have developed Dynamic Tool Destination (DTD), which is tool and destination independent. When DTD is set as the default destination in Galaxy’s job_conf.xml it can replace all tool destinations. DTD matches tools to destinations using rules setup in its own configuration file. If any rules match, it will apply them; if none match, either tool specific or DTD default destination is applied. The benefit is that once job_conf.xml has all of your possible destinations defined, configuring a new tool to use DTD can be done on the fly as it doesn’t require Galaxy to be restarted.
Dynamic Tool Destination is freely available at https://github.com/phac-nml/dynamic-tool-destination