Richard Async usecase
Richard has applications that take a long time to process their data. They are compute-intensive. This includes building phylogenetic trees, microarray statistical analysis R/Maanova and analyzing genotype information to build population structure for crop germplasm Structure and !InStruct.
Currently, these applications are run as command-line jobs. It would be ideal if they could be exposed as services, so that they can be used within pipelines, or launched from off-site. Some preliminary efforts underway with SoapLab?, but the long run-time of these applications prevents naive synchronous methods for exposing them.
- server: http:hpc.irri.cgiar.org for web portal, 3rd party command-line apps, R scripts
- client: custom visualisation clients, probably in Java
- orchestration: taverna, in-house middleware
- licensing restrictions: all their sw is open, would strongly prefer os/free license for code dependencies
Should see: generation challenge program platform