1. Background
Amyotrophic Lateral Sclerosis (ALS) is a fatal neurodegenerative disease characterised by the selective loss of upper and lower motor neurons. Bulk-tissue genomic and transcriptomic analyses have revealed major molecular perturbations across neurons and glial cells, but bulk and cell-type specific insights for epigenetic associations with disease are still lacking. We will leverage large-scale existing DNA methylation aware Oxford Nanopore Technologies (ONT) DNA sequencing data from post-mortem samples of ALS cases and controls, as well as our expertise in epigenetic cell type deconvolution to dissect cell-type-specific epigenetic mechanisms involved in disease.
2. Novelty & Importance
This PhD will develop methods to computationally infer and separate cell-type-specific molecular profiles from bulk Nanopore sequencing of ALS brain samples. By integrating non-negative matrix factorisation (NMF), machine learning, and emerging methylation foundation models, we aim to deconvolve complex bulk datasets into constituent cell-type signals – including motor, glutamatergic, dopaminergic, and glial populations. The work will leverage extensive in-house ONT data from Prof Iacoangeli’s group and neuronal nuclei reference methylomes generated in the Marzi Lab, bridging computational and experimental neuroepigenomics.
3. Aims & Objectives
1) Develop three complementary computational cell type deconvolution frameworks for bulk brain ONT data:
• Non-negative matrix factorization
• Deep learning
• Foundation-model fine-tuning
2) Curate and analyse reference ONT methylomes for major neuronal and glial cell types generated using fluorescence-activated nuclear sorting (FANS).
3) Conduct cross-validation and test the performance of this model on other datasets (including AMBRoSIA, TONiC and the MND Biobank).
4) Apply the models to ALS bulk ONT data to uncover cell-type–specific methylation changes linked to disease progression.