Single Cell Ageing and mtDNA: learning and simulation
Project Description
The aim of this project is o understand how mtDNA mutations accumulate in our cells, and what their effects are, using a mix of simulation, inference and machine learning.
Existing background work
We have been investigating (genetic) variation in cellular power stations (mitochondria) using a mix of ideas from stochastic population processes/statistical genetics, statistical physics, statistical inference/machine learning and control. Our goal is to understand human ageing and to inform therapies to combat disease. We believe this is a particularly exciting area to study for two reasons. Not only does 1) mitochondrial (dys)function have deep connections to therapies for conditions like Parkinsons, Diabetes, ageing and Cancer (and we are working on these) it is 2) a topic that, though poorly understood, is susceptible to the very basic (though mathematically nuanced) models that one constructs in theoretical and mathematical physics/ statistical genetics. This is an area where students can explore a new scientific direction since the medical promise and scientific opportunities easily exceed the number of theorists. Students can develop new evolutionary theory while also contributing to therapeutic design. We place a substantial emphasis on single-cell sequencing and single-cell transcriptomics (these are among the most active areas of modern biomedicine): this involves aspects of machine learning and bioinformatics. We build stochastic models that link to the results of our inference from the transcriptomics. We also link information about cellular state to cellular mutational state using tools from (Bayesian) Machine Learning. The applicant will have an opportunity to participate in a large project with the Cambridge experimental groups of Profs Patrick Chinnery and Maria Spillantini on Parkinson’s disease and mitochondria. Our work will build on our recent paper: ‘Cryptic mitochondrial ageing coincides with mid-late life and is pathophysiologically informative in single cells across tissues and species’ https://www.biorxiv.org/content/10.1101/2023.07.04.547509v1
Main objectives of the project
Exemplar objectives would be to:
Search for evidence for selection of mutated mtDNA
Discover whether epigenetics or mtDNA mutations are better associated with ageing
Develop causal inference tools linking mtDNA mutations to gene-expression
Simulate mutation accumulation in single cells and in tissues
Details of Software/Data Deliverables
Data analysis: machine learning linking mtDNA mutation and gene expression; causal inference linking mutations and genes; (Approximate) Bayesian inference fitting stochastic models for mutations
Simulation: forward-simulating mutation accumulation in cellular populations and simulations.
