Longitudinal Multi-Omic Data Integration for the Study of Evolutionary Dynamics and Awakening in Hormone Dependent Breast Cancer.

点击翻译按钮获取中文摘要

Relapse continues to be a major problem in oestrogen receptor positive (ER) breast cancer. Risk of relapse continues for at least 15 years after endocrine treatment discontinuation, and the recurrence rate remains stable across time. This points towards a subpopulation of cancer cells entering dormancy, acquiring resistance, and then stochastically awakening, leading to relapse. Genomic, transcriptomic and epigenetic profiles at bulk, along with single cell expression and lineage tracing profiles, are available in house from MCF7 and T47D cell lines undergoing dormancy-awakening dynamics in response to endocrine therapy. Previous analyses of this data have highlighted the importance of studying the awakening of dormant cells, and longitudinal multi-omic data integration could unveil new insights into the mechanisms driving this process. Our first aim will be to apply and benchmark established statistical machine learning approaches, including dimensionality reduction, semi-supervised data integration and network analysis. For our second aim, we will develop novel statistical methods for data integration building on the work of the Evangelou Group. In particular, we will consider the addition of a temporal dimension to data integration approaches, an area in multi omic data integration that is currently under-explored. This exploration of temporal data will let us address our third and main aim, understanding the biological and molecular processes underlying the evolution of cancer cells from dormancy to awakening and test hypotheses such as epigenetic erosion.