Somatic and germline mutational processes and the evolution of cancer drives genes

Cancer develops from healthy cells due to mutations in around 600 cancer driver genes. These mutations, along with neutral ones, are generated from DNA mutational processes, creating distinct patterns known as mutational signatures, which are crucial for understanding cancer's etiology. This doctoral thesis explores two main topics: mutational signatures in somatic and germline cells, and the evolution of cancer driver genes. It introduces SigNet, a neural network algorithm that accurately identifies mutational signatures even in samples with low mutation counts, revealing signatures linked to hypoxia, cancer stages, and genomic regions, with a focus on challenging signatures like SBS3, SBS5, and SBS40, and regions such as close to the transcription start sites of genes. The second focus examines if cancer driver genes evolved to minimize mutations that could lead to early-onset cancer. By reconstructing the most recent common ancestor between humans and chimpanzees and simulating germline evolution, the study suggests differences in the evolutionary paths of cancer driver and non-driver genes, potentially linked to DNA repair signatures.