This DPhil project aims to elucidate the mechanisms of RNA splicing within immune cells, focusing on the computational analysis of alternative splicing events and their impact on immune homeostasis and inflammatory responses. By leveraging pre-existing long-read single-cell sequencing data, this research will develop and apply computational methods to analyse splicing patterns, identify key regulatory elements, and integrate multi-omics data to understand the functional consequences of alternative splicing in immune regulation.
Map RNA Splicing in Immune Cells:
Data Utilisation: Utilise pre-existing long-read single-cell RNA sequencing (scCOLOR-seq) data from healthy immune cells to capture the full transcriptome and elucidate splicing regulation mechanisms.
Computational Analysis: Develop and employ bioinformatics pipelines to process sequencing data, focusing on isoform diversity and splicing events. Conduct Differential Isoform Usage (DIU) tests to identify significant splicing changes between conditions.
Elucidate Splicing Regulation Mechanisms:
Data Integration: Integrate RNA-seq data with epigenetic and transcriptomic profiles to identify regulatory elements influencing splicing. Examine the roles of RNA-binding proteins (RBPs) and non-coding RNAs (ncRNAs) in modulating splicing events.
Machine Learning Models: Implement machine learning approaches, such as random forests and autoencoders, to predict regulatory elements and their impact on splicing. Employ feature selection techniques to highlight key regulators.
Identify Causal Features and Construct Predictive Models:
Causal Inference: Use Structural Causal Models (SCMs) to infer causal relationships between splicing regulators and observed splicing patterns. Identify key elements driving splicing modulation under inflammatory conditions.
Predictive Modelling: Develop ensemble machine learning models to predict splicing outcomes based on identified regulatory features. Use Graph Neural Networks (GNNs) to integrate gene interaction data and enhance predictive accuracy.