v0.10.0

Change Log

User-facing changes:

• Add a column DiseaseLocation to the table efo_labels containing anatomical locations associated with each EFO term through the property EFO:0000784 (has_disease_location); if any exist.
  • When a term does not have such a relation, the code tries to find location relations inherited from the most immediate parent(s) in the class hierarchy.
  • Populate DiseaseLocation column with NA when no locations exist.
• Add a table called version_info to document details such as versions of ontologies & the generated DB and the download date of the OpenGWAS metadata used to build the database.
• Add functionality to include additional, user-provided tables in the generated database.
• Add functionality to include additional, user-provided ontologies in the generated database.
• Output a compressed tar.xz archive instead of a .db file.
• Add UBERON ontology tables to the database.
• Update to latest EFO v3.57.0.

Developer-facing changes:

• Fix issue where ontology mappings were being undercounted, because the mapping identifiers we generated were colliding in the event of multiple mappings per trait. Switched to generating random UUIDs for mapping identifiers to avoid such collisions.
• Fix issue with mismatching CURIEs that resulted in mapping count discrepancies.
  • the CURIEs generated by bioregistry were different than those in the SemanticSQL database.
  • updated the bioregistry package version (and some other dependencies) to a more current version that produces CURIEs as expected to match properly.
• Use the SemanticSQL .gz archives instead of the .db files.
  • In the case of EFO, the .db version of EFO was much older than the one in the .gz archive.
• Save efo_labels table with the mapping counts to disk.
• Update query_database.py to work directly with the tar file.
• Updated some dependencies: text2term, pandas, and bioregistry.

v0.9.0

Change Log

User-facing changes:

• Added a table called efo_synonyms to the database, containing synonyms of terms in EFO.
  • The table contains EFO terms represented by their short identifiers in the Subject column (e.g. EFO:0000712) and their synonyms in the Object column (e.g. stroke, stroke disorder, etc.).

Developer-facing changes:

• Added optional input to build script to set a user-given NCBI API key environment variable.
• Added some timers and print statements to track run time of major tasks and overall database build time.

v0.8.0

Change log:

• Upgraded to a more flexible, parameterized database "builder" version, which can be applied to arbitrary metadata sets that either contain ontology mappings (like GWAS Catalog) or that will be mapped per user inputs (using text2term).
• Properly distinguish between direct and inherited mappings, such that the total mapping count is the sum of direct and inherited.
• Mapping scores are rounded to 3 decimal points.
• Citation metadata (title, abstract, journal, etc.) are pulled from PubMed using the PMIDs in the GWAS Catalog metadata.

v0.7.0

Change log:

• Fixed a bug where the database tables were created with (outdated) column names that contained spaces. Table columns are now named using CamelCase, and have data types associated.

v0.6.0

Change log:

• Updated code, data and examples to use CamelCase names for table columns:
  • all efo_* tables have as column names: Subject, Object
  • efo_labels table has additional columns:
    • IRI specifies the full term identifier
    • Direct specifies the number of traits in the metadata directly mapped to the term
    • Inherited specifies the number of traits indirectly mapped to the term (i.e., inherited from a direct mapping)
• Include Disease Ontology (DOID) terms in EFO as potential mapping targets
• Updated mappings with those obtained with text2term v2.3.0.

v0.5.0

Change log:

• Removed deprecated ontology terms in the efo_labels table, which were being brought in from the EFO SemanticSQL DB.
• Added functionality to fetch metadata directly through the OpenGWAS API.
• Updated OpenGWAS metadata to the latest state as of 04/27/2023.

v0.4.0

Change log:

• Added module to compute the counts of direct and inferred mappings to each ontology term.
• Updated database to include mapping counts in the efo_labels table.

v0.3.0

Change log:

• Include EFO term IRIs in the efo_labels table to enable creating hyperlinks.
• Remove duplicates in all tables.
• Remove all but one label row for each ontology term ID (to use term IDs as unique keys).
• Remove leading/trailing whitespace in labels of efo_labels table.
• Modify the output format of example-query results— results are given in TSV files & query parameters used to obtain those results are given in a separate .txt file, which also specifies the number of results returned for each query.

v0.2.0

Change log:

• Added a table with the database cross-references contained in EFO
• Fixed the compact identifiers (CURIEs) in the tables extracted from the SemSQL DB, as in some cases the identifier was the full term IRI (e.g. <http://www.orpha.net/ORDO/Orphanet_1333>). Bioregistry is used to obtain CURIEs from such IRIs.
• Included the results of example queries to facilitate testing.

v0.1.1

Change log:

• Improved documentation throughout.
• Minor refactorings to facilitate debugging.
• Dropped the predicate column across tables as it is unnecessary.