LanDis

Large scale proteomics data has helped clarify the relationship between a disease phenotype and its causes, to a point where it is now clear that a disease is a wider perturbation in the underlying biological networks. Nevertheless, there is relatively little molecular information about the hereditable diseases described in OMIM To put this in perspective, 45% of the known hereditable diseases in OMIM have no known causal genes, and for the remaining ones, the information might not be complete.

On the other hand, there is an abundance of phenotype descriptions that, although not well suited for machine processing, provide comprehensive information on the various diseases. This information reflects the complexity of each disease's molecular foundations and thus quantifying similarity at this level would reflect the closeness at molecular level between two diseases. Accurately quantifying this molecular closeness could shine a light on the discovery of new disease genes and possibly help identify new targets for drugs.

By annotating the diseases in OMIM with the MeSH terms associated to the publications they reference, we are able to obtain high-quality annotations for the diseases. These annotations, in combination with the structure of the MeSH ontologies allows us to obtain a single number that characterises molecular similarity between the diseases with high accuracy.