Data Science Seminar
October 15, 2020
Dominik Glodzik, PhD
Zoom link: https://bit.ly/DSOct15
Abstract: Whole genome sequences contain within them signatures of mutational processes. In particular, some of the mutation signatures relate to impaired DNA-repair in cancer cells. Accurate measurement of mutation signatures reveals the role of DNA-repair deficiencies in etiology and progression of cancer.
We extended the computational methods for analysis of mutation signatures in order to describe patterns of chromosomal rearrangements. In particular, the rearrangement signatures enable the assessment of proficiency of homologous recombination (HR). HRDetect, an algorithm we developed, predicts probability of HR-deficiency, and is based on holistic portrayal of mutational signatures across different classes of somatic mutations. Around 20% of breast cancers contain signatures of HR-deficiency, and this group is wider than the group of carriers of BRCA1/2 mutations. By contrast to adult cancers, pediatric cancers with known DNA-repair defects display variation of mutational signatures, hinting at tissue-specificity of mutational signatures. Finally, in the chromosomally unstable cancers, we identified structural rearrangements, in coding and non-coding regions, that can act as cancer drivers. Altogether, these results indicate that computational assessment of DNA-repair capacity of tumor cells is now possible. The methods will be crucial to understanding of the DNA-repair mechanisms and tissue-specificity of mutational processes.
Bio: Dominik Glodzik received his PhD in Computational Biology from the University of Edinburgh, and held a postdoctoral position at Wellcome Trust Sanger Institute, before moving to a staff scientist position at Memorial Sloan Kettering Cancer Center. Currently he is a Principal Bioinformatician at Repare Therapeutics in Cambridge.