Computational analyses of transcriptomic alterations in prostate and colorectal cancers

Cancer is caused by genetic mutations which give cells growth advantages over their neighboring cells. The infinite number of possible mutation combinations makes no two cancers the same, and there is a pressing need to identify groups of patients that have similar genetic changes that can be targeted for therapy or contribute to a more precise diagnosis and prognosis. The present thesis studied gene mutations in colorectal and prostate cancer. In colorectal cancer, we identified one patient group whose outcome was dependent on how the KRAS gene was pieced together, in a process called splicing. We found another group when we explored the splicing patterns of all 20,000 genes at once. Patients with more splicing changes generally have poorer prognosis than other patients. In prostate cancer, patients are commonly diagnosed with many genetically different tumors within their prostate. The majority of the research on prostate cancer has been based on biopsies from a single tumor per patient, and it is then very important that the most aggressive tumor is sampled. We have identified 16 genes that can be sampled from any tumor in a patient and still be used for prognostication of the patient, no matter if the sampled tumor is the most aggressive one.