Covariance estimation and network reconstruction of multi-omics data (360G-Wellcome-105362_Z_14_Z)

This project will develop and apply novel network-theoretic approaches to study the regulation of RNA transcription on the basis of multiple molecular phenotypes. In particular, we seek to more accurately characterise the regulatory mechanisms that drive haematopoiesis (the process during which stem cells differentiate into more function-specific blood cells) by combining complementary -omic datasets. The methodological aims of the project: i. Develop a flexible Bayesian procedure for the estimation of high-dimensional covariance matrices. ii. Apply such inference tools to reconstruct interaction networks. iii. Extend the model to incorporate multiple sources of information. The applied aims: iv. Improve understanding of haematopoiesis by characterising cell differentiation lineages, specifically the lineages of the Megakaryocyte Erythrocyte Progenitor cell and the Neutrophil lineage of the Granulocyte Monocyte Progenitor cell. This will be done using the method in (iii) by combining multi-omic datasets, including RNA-seq and Chip-seq, as well as the physical network provided by promoter capture Hi-C experiments. v. Reconstruct gene-interaction networks based on single-cell sequencing data, taking into account the strong technical variability that characterises these experiments. We expect this analysis to reveal interactions that have been masked by bulk experiments, where only overall expression over thousands of cells is recorded