DR-050 Discovery and Clincal Characterization of Cancer Driver Mutations in Noncoding regions

Cancer genomes typically harbor a substantial number of somatic mutations. However, only few of these mutations actually contribute to tumor development. To identify these so-called driver mutations, large-scale aggregated sequencing datasets are needed. Over the past decade, the search for driver mutations has been mostly focussed on coding regions. Substantially less is known whether and to which degree mutations in non-coding regions are functionally involved in tumor formation. With the drop in sequencing costs over the past years, it became possible to sequence the entire genomes of large cohorts of cancer patients (e.g. the dataset provided by Hartwig Medical Foundation). However, the avaialalbity of these datasets has not been met with the of algorithms that are particularly designed to the underlying biology of these data yet. The goal of this project is to develop a statistical framework which systematically identifies driver mutations in non-coding regions in aggregated whole-genome sequencing data. For this purpose, we will particularly consider the biological differences between coding an non-coding regions, in order to specifically taylor our algorithm to non-coding regions. Applying this refined approach to ~2500 Whole Genome Sequencing data, we will systematically characterize which mutations in non-coding regions are relevant to oncogenesis.

Eliezer van Allen Dana-Farber Cancer Institute USA