Genetic and molecular variations are closely intertwined; while genetic factors drive phenotypic differences ranging from gene expression to organismal traits, phenotypic variations are the target of evolutionary selection, what eventually results in genetic changes. As technological advances have resulted in high-throughput assays for different molecular dimensions, it has become challenging to turn these large-scale data into meaningful insights and to delineate biological cause and consequence. In this thesis, I use computational modelling to detect and understand biologically meaningful associations between genetic variation and gene expression alterations. First, we use data across 27 human cancer types to probe associations between different genetic factors and gene expression levels. We describe the tumours' regulatory landscape that is highly heterogeneous across cancer types, and quantify the relationship between gene expression and various genetic features that characterise local and global mutational burden as well as distinct mutational processes. Next, we study the relationship between genetic and epigenetic variation and alternative splicing. This analysis extends studies of splicing events in bulk data to variability in splicing between single cells from the same tissue: We analyse DNA methylation and alternative splicing across single cells derived from one human donor to characterise splicing variation and its determinants across genes. Thus, we identify relevant genetic determinants of splicing in induced pluripotent stem cells as well as during their differentiation, and a significant contribution of DNA methylation to splicing variation across cells. Finally, we show how gene expression-mutagenesis screens can be applied to understand complex mutational signatures, using the cancer hallmark of DNA repair deficiency as an example. The molecular cause and consequence of homologous recombination repair deficiency are not yet fully understood. We explore genome-wide molecular aberrations caused by this repair deficiency beyond the few previously known genes. Our preliminary results point towards a genetically dominant effect of BRCA1 mutagenesis. Taken together, this thesis highlights novel dimensions of genotype-phenotype associations in highly heterogeneous molecular datasets. We describe the complex regulatory landscape across human cancer types, as well as molecular alterations and relevant epigenetic effects in differentiating pluripotent stem cells.