Human Disease Ontology 2018 update: classification, content and workflow expansion

Abstract

The Human Disease Ontology (DO) (http://www.disease-ontology.org), database has undergone significant expansion in the past three years. The DO disease classification includes specific formal semantic rules to express meaningful disease models and has expanded from a single asserted classification to include multiple-inferred mechanistic disease classifications, thus providing novel perspectives on related diseases. Expansion of disease terms, alternative anatomy, cell type and genetic disease classifications and workflow automation highlight the updates for the DO since 2015. The enhanced breadth and depth of the DO's knowledgebase has expanded the DO's utility for exploring the multi-etiology of human disease, thus improving the capture and communication of health-related data across biomedical databases, bioinformatics tools, genomic and cancer resources and demonstrated by a 6.6× growth in DO's user community since 2015. The DO's continual integration of human disease knowledge, evidenced by the more than 200 SVN/GitHub releases/revisions, since previously reported in our DO 2015 NAR paper, includes the addition of 2650 new disease terms, a 30% increase of textual definitions, and an expanding suite of disease classification hierarchies constructed through defined logical axioms.

Figure 1.

Advanced Boolean Searches. AND/OR searches of any of the DO datatypes (Name, Synonym, Definition, SubSet, DOID, Alternate ID, Xrefs) enable complex data queries of the DO Knowledgebase. For example, a search of Xrefs: OMIM, Name: Parkinson, Subset: DO_rare_slim – identifies all DO disease terms that include a cross-reference (Xref) to OMIM, where the disease name includes ‘Parkinson’ and where the disease is included in the DO_rare_slim – rare disease category. This query in DO, returns seven disease terms.

Figure 2.

An example DO SPARQL: doid-report.rq.

Figure 3.

Example areas of term expansion: (A) hematopoietic system diseases and (B) inherited metabolic disorders. Blue arrow indicates new DO terms.

Figure 4.

Skin disease logical axioms, define inferred disease parents. Integration of a SubClass Of logical axiom for ‘ichthyosis vulgaris’ and ‘autosomal dominant cutis laxa’ [BOLD] (‘has material basis in’ some ‘autosomal dominant inheritance’) and an ‘Equivalent To logical axiom for ‘autosomal dominant disease’ (disease and (‘has material basis in’ some ‘autosomal dominant inheritance’)), where ‘autosomal dominant inheritance is from the Genotype Ontology (http://www.obofoundry.org/ontology/geno.html) creates ‘inferred’ child to parent DO relationships, thus both skin diseases are defined as inferred child terms of DO’s ‘autosomal dominant disease’.