Section: New Results

Alcyone system for repeatable e-science

One of Pleiade 's goals is to assist scientific users in deploying analysis software in their desktop environments. Increasingly, this is not a question of installing software packages locally, but of building bespoke environments that comprise many cooperating software tools. A typical example is a local Galaxy instance, communicating with a project-specific database that is shared with visualization and analysis tools, and cooperating with an electronic notebook such as Jupyter. In order to foster repeatable science, the configuration of each such environment should be reliably recorded, in a way that allows it to be redeployed in the future or shared with a colleague.

Pleiade 's Alcyone system provides a mechanism for specifying and deploying software environments for scientific users in bioinformatics and biodiversity. Alcyone offers three facilities:

1. A specification using configuration-by-convention style, combining specification files in YAML format and raw data files.

2. A collection of Dockerized services that can be chosen in the specification.

3. A deployment system that compiles the specification into a master container image, which orchestrates the deployment and management of the service containers.

The user's environment is fully specified in files that can be archived and shared, allowing future reuse. The use of Docker containers guarantees that future deployments run exactly as before, since the precise versions of the service containers are recorded.

Furthermore, Alcyone specifications are files, that can be managed by the Git source code control system. Different versions of the environment, including different analysis pipelines and intermediate results, are stored in the Git history and any version can be resurrected and deployed. Git branches can also be used to share configurations between users in the same lab.

Alcyone is being tested internally by Pleiade and is undergoing intense development. Existing service containers are Pleiade 's Magus knowledge base, Magecal gene prediction pipeline, and Mimoza metabolic network explorer ; as well as third-party tools Galaxy, Gbrowse, and Jbrowse.