Week 6: Troubleshooting ECSO and Reevaluating Arctic ECSO

This week we had some major set backs. After speaking with Margaret O’Brian last week I realized that the version of ECSO I was opening on Protégé 5 was incomplete. The version of ECSO I was using did not properly import the external ontologies. Many of the terms I was creating and annotating therefore were unnecessary. For example, all of the ecosystem terms I created last week are synonymous with the imported ENVO biome classes. From the original terminology list of Arctic Data Center searches 49 of 136 are already present in ECSO.

In light of this, we decided not to move forward with the Arctic ECSO development until ECSO issues are resolved. The issue of the external ontology imports is the top priority since without those imports we cannot further expand the ontology. In order to import external ontologies the Minimum Information to Reference an External Ontology Term (MIREOT) format is standard for ontology design. In order to follow this method a web interface Ontofox has previously been utilized by the creators of ECSO to import terms in the MIREOT format. The Ontofox web-based system (read here for more detailed information)  is a extraction tool where ontologists can input a search file that Ontofox uses to gather information from external ontologies using a dedicated SPARQL-based ontology term retrieval algorithm and creates a new owl file of those user-selected terms. In order to begin addressing the issue of external import errors I completed the Ontofox tutorial. As a test I imported the entire ENVO ontology into ESCO. This duplicated the ENVO terms that had previously been imported and created the new classes that had not previously been imported. upon reopening this saved ontology the import functioned properly. While this is because the ontology reference information was stored in the catalogue file xml on my computer it may be that the old ontology is not 100% compatible with Protege 5 as it was built in Protege 4 ? I am just speculating at this point. While working on this issue  I came across several CHEBI terms which had been imported from ENVO apparently and were  missing most of there annotation information. Now it is possible that many of these “naked” terms I am seeing in ECSO were imported without the necessary annotation properties established by Sara Lafia and others. But, I suspect that the issue is with the imports themselves. There are several Arctic ECSO terms, which are available in ENVO and CHEBI that need to be added to ECSO. I discovered these new terms can be imported into the open ontology window and merged with the existing external ontology which will be useful when we get there. So, in short the external imports are broken in ECSO and directly importing them isn’t difficult on the user level but they should be automatic for users.

Another issue with ECSO I have been working on is the issue of scientific label versus ontology label. Margaret asked me to copy all of the rdfs labels to the annotation property skos hidden label. So that she and Mark can begin changing the ontology labels to scientific labels.  I am trying to figure out a way to do this “batch” style without manually annotating each individual class, since there are 4,259 of them!

Leave a Reply

Your email address will not be published.