Modelling gene expression dynamics with Gaussian process inference

Abstract

Gaussian process (GP) inference provides a flexible nonparametric probabilistic modelling framework. We present examples of GP inference applied to time series gene expression data and for single-cell high-dimensional ‘snapshot’ expression data. We provide a brief overview of GP inference and show how GPs can be used to identify dynamic genes, infer degradation rates, model replicated and clustered time series, model stochastic single-cell dynamics, and model perturbations or branching in time series data. In the case of single-cell expression data we present a scalable implementation of the Gaussian process latent variable model, which can be used for dimensionality reduction and pseudo-time inference from single-cell RNA-sequencing data. We also present a recent approach to inference of branching dynamics in single-cell data. To scale up inference in these applications we use sparse variational Bayesian inference algorithms to deal with large matrix inversions and intractable likelihood functions.

Publication
Handbook of Statistical Genomics