THESIS DEFENSE — PUBLIC SEMINAR - Bioinformatic Analysis of Physical Mutation Signatures in Tumors to Infer Carcinogenic Processes and Exposures
Amphitheatre, Level 2, Duke-NUS
IBM PhD PROGRAM (INTAKE 2012)
The tumor somatic mutation catalog is the combination of mutation signatures left by all the mutational processes that operate on the cancer cells. This thesis advanced knowledge of mutation signatures and techniques for understanding them in several areas. First, going beyond the single-nucleotide-variant mutation patterns used in conventional mutation signature analysis, I used small insertions and deletions detected in next-generation sequencing to identify a signature of mismatch repair deficiency. Second, current methods for estimating the activities of various signatures in a given tumor tend to find every signature in every tumor. To overcome these limitations, I used additional criteria to estimate mutation burdens ascribable to the exogenous mutagen aristolochic acid. Third, going beyond in-silico data analysis to experimental data, I extracted a ‘pure aflatoxin B1’ signature from aflatoxin B1-treated cell lines and mice, and used this signature as a reference to identify aflatoxin exposed liver cancers. In conclusion, the thesis exemplifies approaches for signature anaylsis that extend the standard methods in important directions, and represents substantial contributions to our understanding of mutational signatures and methods for analyzing them.
Thesis Advisor: Prof Steve Rozen