DR-413 Identifying non-coding driver regions in cancer using Machine Learning

Patients with some types of cancer can have their tumor DNA sequenced to identify tumor variants. The identification of specific types of variants can alert the doctor to recommend targeted therapies that are known to work best for that specific type of tumor. Currently, these sequencing studies focus on regions within the genes that ‘code’ for proteins. Little is known about the utility of all other ‘non-coding’ regions adjacent to the genes or regions that are regulating the genes. The main bottleneck of this type of research is the number of profiles of tumor variants available for study. By compiling information of tumor variants observed in HMF participants we aim to identify non-coding regions that are potentially driving cancer. We believe developing methods to identify ‘non-coding’ driver regions will serve to expand the information that can be derived from the sequencing of tumors resulting in better care for patients with cancer.