SPEAKER:  Dr. Camila de Souza, Assistant Professor in Data Science at Western University

SPEAKER'S BIO:  Dr. de Souza’s research program consists of developing new statistical methods to analyze large and complex data structures arising from various areas in the Natural Sciences, Health and Engineering, such as cancer genomics, molecular biology and electrical engineering. Dr. de Souza received her PhD in Statistics at the University of British Columbia and was a postdoctoral fellow at the BC Cancer Research Centre working on statistical methods for high-dimensional sequencing data. Dr. de Souza is originally from Brazil where she received both her Bachelor's and Master's degrees at the University of Campinas.

ABSTRACT:

In this talk I will present Epiclomal, a probabilistic method to cluster sparse CpG-based DNA methylation data from single-cell whole genome bisulfite sequencing (sc-WGBS). Our approach is based on a hierarchical mixture model, which pools information from observed data across all cells and neighbouring CpGs to infer the cell-specific cluster assignments and their corresponding hidden methylation profiles. Using synthetic and published single-cell CpG datasets we show that Epiclomal outperforms non-probabilistic methods and is able to handle the inherent missing data feature which dominates sc-WGBS. Using a novel sc-WGBS dataset from breast cancer xenografts, we show that Epiclomal discovers sub-clonal patterns of methylation in aneuploid tumour genomes, thus defining epiclones. We show that epiclones may transcend copy number determined clonal lineages, thus opening this important form of clonal analysis in cancer. A preprint is available at bioRxiv (https://doi.org/10.1101/414482 [1]).