In September I successfully defended my DPhil at Oxford on “Probabilistic Modelling of Genomic Trajectories”. This tackles the single-cell pseudotime problem from a (mostly Bayesian) statistical latent variable approach.
The entire thesis is available (39Mb) or broken down by chapter below. If you find any typos please let me know (but don’t tell Oxford).
- Chapter 1 An introduction to single-cell RNA-seq analysis, the pseudotime estimation problem, and statistical latent variable modelling
- Chapter 2 Examining uncertainty in single-cell pseudotime using Gaussian Process Latent Variable Models
- Chapter 3 Learning single-cell pseudotime using Bayesian nonlinear factor analysis models
- Chapter 4 Inferring bifurcations using a Bayesian hierarchical mixture of factor analysers
- Chapter 5 introduces covariate-adjusted latent variable models for learning latent variables from heterogeneous samples (e.g. with different genetic/environmental backgrounds)