Towards a more molecular taxonomy of disease

• DOI: 10.7490/f1000research.1112692.1
• Publication Date: 2016-07-25
• Authors: Jisoo Park; Benjamin Hescott; Donna K. Slonim
• Keywords: Tree (set theory); Plant disease; Complex disease

Disease taxonomies have been designed for many applications, but they tend not to fully incorporate the growing amount of molecular-level knowledge disease processes, inhibiting research efforts. Understanding degree which we can infer relationships from molecular data alone may yield insights into how ultimately construct more modern that integrate both physiological and information. We introduce a new technique call Parent Promotion hierarchical between terms using disease-gene data. compare this with an established ontology inference method (CliXO) minimum weight spanning tree approach. Because there is no gold standard taxonomy available, our inferred hierarchies Medical Subject Headings (MeSH) category C forest diseases subnetworks Ontology (DO). This comparison provides about algorithms, choices evaluation metrics, existing content various MeSH DO. Our results suggest performs well in most cases. Performance across trees also correlated methods. Specifically, are consistent those smaller than larger ones, some notable exceptions correlate higher MeSH. experiments provide learning genes alone. Future work should explore prospect term discovery best anatomical clinical knowledge. study nonetheless suggests gene information has potential form important part foundation future representations landscape.