Purpose

This is an ontology to annotate omnibenchmark runs.

We provide three components:

• the ontology formalization (turtle)
• an example knowledgebase (turtle) with a simplified setup:
  • one data module
  • two filtering modules
  • one method module, run with different parameters
• another (more realistic) example, the the iris-example knowledgebase somewhat resembling the omnibenchmark named iris-example
• some SPARQL queries, including:
  • get_deliverables_by_activity.sparql
  • get_deliverables_by_epoch.sparql
  • get_deliverables.sparql
  • get_file_provenances_one_hop.sparql
  • get_file_provenances_recursive.sparql
  • get_modules.sparql

To generate these triples and send them to the endpoint: adopt and reuse epoch.py from omni-cli (it's mainly python native templating).

Design philosophy

• Keep it deliverable-centric (file-centric), because bioinformatics workflows typically are.
• Reduce overlaps to renku's ontology but keep in mind future cross-queries; we can add federated queries and/or merge graphs by matching literals (i.e. plan ids or activity ids).
• Make use of gitLFS / S3 buckets identifiers, to streamline dataset retrieval in a renku-free manner.
• Make use of module names and file names, to perhaps switch to renku-free command-line workflows at some point.
• Keep versions/epochs explicit.

Ontology layout

Layout generated by WebVOWL 1.1.7.

Requirements

• Specs

To remember

• Missing git hook/populate the endpoint on push
• To be linked to the container/codebase tagging/branching system
• (Sept 2023) Renku's named key:value parameters are named parameterMaps and only apply to composite plans, not to plans: source
• renkulab API queries (to potentially replace with our custom queries)