Reconstructing and analysing cellular states, space and time from gene expression profiles of many cells and single cells.

Genome-wide gene expression profiling is a fast, cheap and standardised analysis that provides a high dimensional measurement of the state of a biological sample. In this review we describe computational methods that can be applied to identify and interpret sources of variance in gene expression in whole organisms, organs, tissues or single cells. This allows the identification of constituent cell types and states in complex mixtures, the reconstruction of temporal trajectories of development, differentiation and progression, and the reconstruction of spatial patterning. When applied to genetically variable samples, these methods allow the efficient investigation of how genetic variation influences gene expression and biological processes in space and time.