Fig 5. Validating inferred has_indication associations.

All 18,889 has_indication associations captured in our integrated network were extracted. These associations were used as a means of validating the ability of our approach to identify known has_indication associations. Note: For each disease category (ALL, C04 and C10) the set of known indications were pruned to only include those containing drugs included in the inferences made by our approach (totalling 17,883). Mapping was done using a Sim value of 0.633, this is equivalent to a distance of two nodes in the MeSH hierarchy.