Biomappings: Community curation of mappings between biomedical entities

Many related biomedical resources propose their own identifiers for genes, proteins, chemicals, biological processes, and other entities of biological interest. The integration of data and knowledge fundamentally relies on mappings between equivalent entities across resources. While some maintain and distribute mappings to external namespaces (e.g. HGNC provides mappings to Entrez Gene identifiers), there exist systematic gaps in the availability of mappings between widely used resources. Automated approaches including lexical and structural alignment have been used to find missing mappings, but most do not store important metadata like mapping confidence nor important curation artifacts including proposed mappings curated to be (nontrivially) incorrect, nor provide interfaces for curating (reviewing, and confirming or rejecting) predicted mappings. We introduce Biomappings, an open repository for making expert-curated mappings (currently 5,700), and predicted mappings (currently 28,000) available with their associated metadata in an intuitive tab-delimited format. It is licensed under the permissive CC0 license to encourage community contributions and restriction-free integration back into primary resources. Biomappings provides a web-based interface for curating predictions and adding manually constructed mappings, a Python package for interacting with the mappings, and several workflow examples for generating new mappings. We applied the Biomappings curation workflow to missing mappings between the Medical Subject Headings and several other ontologies including the Disease Ontology, ChEBI, and the Gene Ontology. We also used Biomappings to curate an exhaustive set of proposed mappings (constructed automatically based on lexical overlap) between entries representing cancer cell lines across three previously unmapped resources: the Cancer Cell Line Encyclopedia, the Experimental Factor Ontology, and Cellosaurus. All data and code are available at https://github.com/biopragmatics/biomappings .