Somatic mutations in cancer genomes are dominated by non-coding variants, yet robust identification of non-coding driver mutations and characterization of underlying mutational processes remain challenging. The aim of this project is to continue our established work on the detection and interpretation of non-coding driver mutations and the study of mutational processes in cancer using large-scale whole-genome datasets. We will analyse publicly available cancer genomics data, primarily from The Cancer Genome Atlas (TCGA) and other international consortia, focusing on whole-genome sequencing data and associated annotations. The project will apply and further develop computational methods for identifying signals of positive selection in non-coding regions, including regulatory elements and other functional genomic features, as well as for extracting and interpreting mutational signatures and related processes shaping cancer genomes. Analyses will include variant annotation, background mutation rate modelling, statistical driver detection, and mutational signature inference across multiple tumour types. The project builds directly on existing pipelines and analytical frameworks routinely used by the research group. All data analysed are generated elsewhere and accessed under relevant controlled-access agreements where applicable. We have approval via dbGap for using restricted access data from TCGA. The project does not involve the generation of new sequencing data. Results are expected to improve our understanding of how non-coding alterations and mutational processes contribute to cancer development and heterogeneity.