In cancer, most somatic mutation are so-called passenger mutations with no functional impact on disease development. Only a subset of the mutations act as drivers and are responsible for tumor growth and progression. Identifying patient-specific driver mutations is one of the main challenges in precision oncology. Existing computational methods to identify patient-specific cancer drivers either integrate mutation and expression data with a single aggregated gene interaction network or use personalized gene interaction networks without considering mutation data. As yet, no methods making use of both patient-specific mutation and gene network data exist. For this reason, we developed PARADYS (PAtient-specific RAnking of genes driving DYSregulation in cancer), a computational tool for personalized detection and impact scoring of genes driving dysregulation in cancer (https://github.com/bionetslab/PARADYS). On several cancer cohorts, we show that PARADYS is able to make robust predictions by integrating patient-specific mutation and dysregulation data. Furthermore, PARADYS’ driver predictions allow for patient stratification into functionally coherent and biologically distinct subgroups. In particular, a case study in prostate cancer reveals a subgroup of patients with infiltration of dendritic cells in the tumor micro-environment and unexpectedly high survival times, highlighting the potential of dendritic cell therapy in prostate cancer.