Aim 1.3 Extend and improve the semantic infrastructure required for phenotype ontology development

matentzn commented 2 years ago

The effectiveness of term templates for automated classification and alignment depends on a broader semantic infrastructure of ontologies they reference including ontologies of relations, phenotypic qualities, species independent anatomy (Uberon), cell types (CL), and biological processes (GO). We already work closely with the developers of these and many other ontologies. Where modification of these ontologies is required to improve the effectiveness of term templates, we will either update them directly where we have access, or work closely with their developers to implement the required changes.

Key people

@dosumis
@rays22
@matentzn
@cmungall
Many more

Key results

Spread the use of ODK release infrastructure to dependent ontologies. Vastly overhaul key release pipelines such as Uberon.
Develop slims (PRO, CHEBI) to improve the resilience of ontology release pipelines

Comments

This goal is never really finished, but we have done a lot to satisfy it. That is what I left it at 99%, because I don't think it needs to be mentioned prioritised much any more.

sagehrke commented 1 year ago

David OS heads this subaim.

matentzn commented 1 year ago

Key results 2022/23

Huge push on cleaning Uberon dependency issues by implementing a base-module solution - this was an enormous feat and helped unblock a lot of issues related to circular dependencies
Many discussions about cleaning of NBO as source for behaviour terms - decided to move phenotype terms out into OBA which will provide a major level of clarity in the scope discussion
All major dependencies have been migrated to ODK, the last one being RO
Cleaning ontologies up, like making sure deprecated classes actually exist and are not buried behind alt_id.
Integrated HPOA annotations into the HPO process which ties the two closer together (advantages for versioning and QC)
Worked with external collaborators on a hierarchy of mapping predicates that can be used for cross species mappings. These will be important for the dissemination of Monarch cross-species mappings
Ongoing updates/releases of all source ontologies
Developed automated actions that can update the imports of ontologies without requiring any local/manual work
OBA Paper published: https://link.springer.com/article/10.1007/s00335-023-09992-1
Integration into Laboratory domain (LOINC) by decomposing LOINC into constituents which are mapped to OBA traits. In Q2, LOINC patterns were identified and mapped to OBA. We will see the first batch of finalised mappings in Q3.
Started working on an entire reboot of the uPheno 2 pipeline, including its integration into OLS4
Included SWISSLIPID lipids in OBA. This is part of a larger effort to created traits in OBA for GWAS, which we want to explore for variant prioritisation.
Supervised the creation of an initial pipeline for the generation of semantic similarity profiles.

monarch-initiative / phenomics_first_resource