Computational systems approaches for reconstructing trans-regulatory networks for decoding cellular mechanisms

Abstract

Multicellular organisms rely on complex trans-regulatory networks (TRNs) to coordinate cellular functionality. Understanding TRNs has profound implications for developmental biology, disease mechanisms, and precision medicine. This thesis presents computational frameworks to explore TRNs at bulk and single-cell resolutions, facilitating a deeper understanding of cellular identity and cell-fate decisions.

My exploration of TRNs begins with an investigation of phosphorylation-based cell signalling (phospho-signalling). I reviewed existing methods and contributed to the development of PhosR, a tool suite for analysing large-scale phosphoproteomics data. Specifically, I worked on the identification of stably phosphorylated sites (SPSs) for data normalisation and batch effect removal. Following this, I developed a statistical framework to identify SPSs across diverse human cell/tissue types. My characterisation of SPSs revealed their evolutionary conservation and functional importance, with potential implications for cancer progression. Additionally, I designed 'SnapKin', an ensemble deep learning model for kinase-substrate prediction, a crucial step in reconstructing phospho-signalling networks.

To link phospho-signalling with downstream regulation of cell identity and fate, I conducted a multi-omic profiling study of myogenesis as a model system to reconstruct the TRNs underlying this process. By integrating multimodal data, I uncovered regulatory networks that govern muscle development and identified the transcription factor Nuclear Factor 1 X-type (NFIX) as a key regulator of myogenesis downstream of MAPK signalling. I further developed 'Refate', a computational framework for identifying key regulators of cellular conversion using large-scale single-cell multimodal omics data and predicting chemical compounds to modulate TRNs.

Taken together, this thesis advances methodologies for reconstructing TRNs, providing insights into cell identity and cell-fate decisions.