Hannah Carter
Assistant Professor of Medicine
University of California, San Diego
Seminar Information
Decades of research have established clear evidence that genetic factors contribute to cancer predisposition, however the genetic risk loci uncovered to date through familial and genome-wide association studies (GWAS) only account for a small fraction of the expected genetic risk. These studies have been designed around the traditional classification of tumors by tissue type or subtype, however recent efforts to exhaustively catalog the set of molecular aberrations in tumor cell genomes suggest an alternative grouping of tumors by molecular profile. We hypothesized that germline cancer risk could in fact be a predisposition to acquire specific somatic "gateway" mutations capable of initiating tumorigenesis. To investigate this possibility we looked for correlation among genetic loci and somatic mutations using data available from The Cancer Genome Atlas. Taking this approach, we found numerous candidate genetic loci where that are strongly correlated with somatic mutation or copy number status in known cancer genes. We were able to validate a number of these loci in a second independent cohort of TCGA samples. Surprisingly, most of the implicated interactions are distant in the genome, suggesting that the architecture of cancer predisposition may be more complex than previously believed.
The primary focus of my research is to develop and apply quantitative methods for genotype-phenotype mapping, particularly in the context of personalized tumor genome analysis. I am particularly interested in integrating information from sequence, structure and knowledge of higher order interactions at the intracellular level to better understand the consequences of genetic variation. I have previously developed machine-learning based methods, CHASM and VEST to discriminate causal "driver" mutations from neutral "passenger" mutations in tumor genome sequencing data. CHASM in particular has already been widely adopted by the cancer research community, and has been integrated into the BROAD Firehose pipeline used to annotate all samples processed by TCGA. I have also developed methods to use networks to identify similar patient groups on the basis of their somatic mutation profiles.