Ontology: when integrating EDAM, force use of EDAM as SAB

[AlanSimmons]

Issue

The ontology generation framework uses PheKnowLator to convert OWL files to OWLNETS format. The PheKnowLator script expects IRIs for nodes to be in a format "http://purl.obolibrary.org//_--e.g.,

http://purl.obolibrary.org/obo/GO_0072179

The script resolves the SAB and code for the concept.

EDAM node IRIs do not conform to the expected format. Examples:

• http://edamontology.org/format_3750
• http://edamontology.org/operation_0287

For EDAM codes, the PheKnowLator script assigns the EDAM subdomains (e.g., format, operation) instead of EDAM.

Based on feedback from the Data Distillery project, we think that this might be confusing--i.e., users would expect EDAM as the SAB.

Solution

1. Use EDAM as the SAB for EDAM codes.
2. Include the EDAM subdomain in the codes.

Example, the node relating to IRI http://edamontology.org/format_3750 should be "EDAM format_370" instead of "format 370".

This will entail either building a custom OWL-OWLNETS converter for EDAM (e.g., a wrapper for PheKnowLator) or a post-conversion fix as part of ingestion. The latter is likely easier.

[AlanSimmons]

The script now formats EDAM node IRIs so that they use EDAM as the SAB. I validated by generating a new knowledge graph.