Abstract
Association studies of complex diseases with a strong genetic component aim to unveil the genomic variants involved in the disease onset mechanism. The knowledge of these variants allows the design of mathematical models for diagnosis and prognosis, and their application in precision or personalized medicine. Unfortunately, in the case of epilepsy, it appears that a major role in the disease onset is played by rare or even private genomic variants. Therefore traditional GWAS failed so far to explain epileptogenesis, even when large cohorts (thousands of patients) were involved. In addition to the typical gene-only association search, here I will discuss a general framework merging domain, pathway and protein-protein interaction data from heterogeneous data sources. By mining external knowledge bases it is possible to boost the association results, retrieve a network of interlaced elements, and design more efficient prediction models.