The field of epigenetics deals with heritable changes in gene function while underlying DNA stays the same. Some examples of epigenetic modifications are DNA methylation and histone modification. There are several epigenetic processes like bookmarking and X chromosome inactivation, but throughout the PhD we aim to focus on gene silencing also termed as regulation of gene expression.
New technologies have achieved gathering epigentic and gene regulating information from same type of cells, which remove potential interactions between different cells. These data are very noisy but we can exploit high number of cell replication to infer meaningful results. During the PhD our aim is to develop novel statistical methods, which carefully address corrupting mechanisms encountered on single cell data, tailored for existing and future technologies for various epigenetic marks like methylation, expression or accessibility. Since single cell technologies are relatively new and research has mainly focused on understanding each epigenetic mark separately. By jointly modeling them, we expect to improve our understanding of impact of epigenetics on gene silencing. Furthermore we would like to identify important genes which can help practitioners in the field carefully design their experiments. We hope to achieve that using Bayesian modelling with various levels of hierarchy. This approach maintains interpretability but is complex enough to address type of noise. While the methodology we aim to develop as part of this PhD is tailored to a specific data science problem, it could be used for inference of highly corrupted data.