Understanding relationships between diseases, such as comorbidities, has important socio-economic implications, ranging from clinical study design to health care planning. Most studies characterize disease comorbidity using shared genetic origins, ignoring pathway-based commonalities between diseases. In this study, we define the disease pathways using an interactome-based extension of known disease-genes and introduce several measures of functional overlap. The analysis reveals 206 significant links among 94 diseases, giving rise to a highly clustered disease association network. We observe that around 95% of the links in the disease network, though not identified by genetic overlap, are discovered by functional overlap. This disease network portraits rheumatoid arthritis, asthma, atherosclerosis, pulmonary diseases and Crohn’s disease as hubs and thus pointing to common inflammatory processes underlying disease pathophysiology. We identify several described associations such as the inverse comorbidity relationship between Alzheimer’s disease and neoplasms. Furthermore, we investigate the disruptions in protein interactions by mapping mutations onto the domains involved in the interaction, suggesting hypotheses on the causal link between diseases. Finally, we provide several proof-of-principle examples in which we model the effect of the mutation and the change of the association strength, which could explain the observed comorbidity between diseases caused by the same genetic alterations.

Authors acknowledge support of Spanish Ministry of Economy MINECO grant BIO2014-57518-R, Instituto Carlos III (ISCIII) FEDER grants CP10/00524 and PI13/00082, and EU H2020 Programme 2014-2020 under grant agreement no. 634143 (MedBioinformatics). CRP is supported by Spanish Ministry of Economy and Competitiveness FPI fellowship (BES2013063354). EG is supported by EU cofounded AGAUR Beatriu de Pinos fellowship from Government of Catalunya. The Research Programme on Biomedical Informatics (GRIB) is member of the Spanish National Bioinformatics Institute (INB) and PRB2-ISCIII, supported by grant PT13/0001/0023 of the PE I+D+I 2013-2016 funded by ISCII-FEDER.