DO Spanish: enhancing accessibility via a standardized workflow for translating ontology and website content

The wealth of biomedical data, though abundant, is not truly available for knowledge discovery across underrepresented linguistic minorities. Here, we report on our efforts to address these disparities for the Spanish-speaking community, through the development of a structured, re-usable translation workflow. The Disease Ontology Knowledgebase (DO-KB, https://www.disease-ontology.org/), a Global Core Biodata Resource, provides a standardized etiology-based classification of the breadth of human diseases, their mechanisms and features. Addressing disparities in disease-data representation, we now provide a Spanish version of the DO-KB website and are in the process of producing a Spanish-translated set of ontology data files. This work describes a structured workflow to facilitate disease-knowledge discovery and to provide the ontology community with a reproducible process for integrating other languages within ontological representations. In order to fully translate the DO in short order, we devised a two-step protocol. We contracted a Texas-based medical Spanish translation service to conduct the translations for DO’s website content (30,883 words). Following validation of the translations (by a DO team member - native Spanish speaker), we submitted the ontology content for translation (disease names, definitions and synonyms) for the DO branches and other DO-KB ontologies. We have established a translation workflow (Google translate, alignment consensus assessment, manual review) in order to provide validated translations. For example, utilizing the 533 DO disease terms (labels) in the physical disorders branch to determine our definition of ‘exact’ matches to allow for variation in case-sensitivity, spacing, punctuation or number formatting (arabic vs roman numeral) differences. This assessment will enable us to establish cutoff criteria for the remainder of the ontology.  We will present the full standardized workflow, update cycle and the release process of DO_Spanish (.owl, .obo and .json) files.  The workflow and data files will be made available (CC0) in the DO’s GitHub repository.