A joint project between fellow NCRIS capabilities, TERN and the Australian National Data Service (ANDS), is not only promoting reusable data, but also reproducible and reusable analysis processes.
Sound science, and the decisions we make based on it, relies on reproducibility. The ability for another researcher to replicate analyses to validate theories and produce complimentary or entirely new science is essential – something that the great thinkers like Aristotle became well aware of when coming up with scientific method. Reproducible science increases a project’s transparency and is essential in facilitating auditing and independent review.
A screen shot of a virtual machine available from web browser through CoESRA that will provide an outlet for people to compose, execute and share their experiments via a cloud-based virtual environment.
|A gallery of tree ferns at the feet of Mountain Ash. A case study on this ecosystem will bring together data and analysis required to apply IUCN Red List of Ecosystems criteria to the Mountain Ash forest ecosystem and make a repeatable workflow for the ecosystem assessment of Mountain Ash forests in the central highlands of Victoria. (Photo courtesy of ccdoh1/flickr CC BY-NC-ND 2.0).|
Reproducibility not only relies on the reuse of data, something which is at the heart of TERN, but also the reuse of ‘meta-analyses’. CoESRA will facilitate reusable simulation analyses by creating an outlet for people to compose, execute and share their experiments in a cloud-based virtual environment.
Specifically, it aims to connect existing TERN infrastructure and other data services running from the National eResearch Collaboration Tools and Resources (NeCTAR) and stored in Research Data Storage Infrastructure (RDSI) with the tools for analysis and manipulation to develop a CoESRA (collaborative environment for ecosystem science research and analysis) virtual experiment environment.
The CoESRA is a cloud-based web-platform that allows researchers to perform complex analyses on a virtual machine without having to set up the experiment from scratch and worry about having enough resources to run the analysis. The virtual machine will be available via a web browser for anybody from Australian Access Federation (AAF) associated organisations.
The CoESRA will provide an opportunity not only to re-use data but also tools for data manipulations, scripts for data visualisation and algorithms for analysis processes. Once a specific analysis is conducted, the entire process chain can be stored and shared with other scientists improving the reproducibility and repeatability of the experiments. Finally, the meta-analysis can be published online at myExperiment and be discoverable via Research Data Australia (RDA).
And, it’s not just researchers that will benefit from the project. The community will gain access to data streams, tools and often hidden ‘pipeline’ processes to leverage further knowledge about ecosystem science experiments.
To showcase the usability of CoESRA, the project will use two real-life case studies: 1) animal conservation and management plan; and 2) ecosystem assessment of Mountain Ash forest.
The first case study will link conservation planning software Marxan and animal tracking information to increase the analytical power of these ecological tools and create a repeatable and reusable workflow for subsequent studies of habitat conservation.
The second case study will bring together data and analysis required to apply IUCN Red List of Ecosystems criteria to the Mountain Ash forest ecosystem and make a repeatable (e.g. re-run in the future with updated time windows and updated ecological parameters) workflow for the ecosystem assessment of Mountain Ash forests in the central highlands of Victoria. The workflow will also enable researchers to re-run the entire assessment with additional data for certain criteria.
Not only will this exciting project reduce the overheads of setting up environmental analyses but will also provide a great benefit to the science community via the creation of a platform for re-usable, repeatable and above all reproducible scientific analyses.
The CoESRA system is currently under development and will be released by the middle of 2015.
- For more information on the CoESRA project please contact TERN’s Siddeswara Guru.
Published in TERN newsletter March 2015