Maximum parsimony analysis of gene expression profiles permits the reconstruction of developmental cell lineage trees

Spatiotemporal control of gene expression lies at the heart of generating several hundred distinct cell types required for the development of higher order animals. Different cell types within complex organs are often characterised by means of genome-wide gene expression profiling, but analogous information for early developmental as well as adult stem and progenitor cells is largely missing because their identity is commonly unknown or they are present in prohibitively small numbers. Here we show that maximum parsimony approaches previously used to reconstruct evolutionary trees from gene content of extant species can be adapted to reconstruct cellular hierarchies both during development and steady state homeostasis of complex mammalian tissues. Using haematopoiesis as a model, we show that developmental trees reconstructed from expression profiles of mature cells are not only consistent with current experimentally validated trees but also have predictive value in determining progenitor cell specific transcriptional programmes and lineage determining transcription factors. Subsequent analysis across diverse developmental systems such as neuronal development and endoderm organogenesis demonstrated that maximum parsimony-based reconstruction of developmental trees represents a widely applicable approach to infer developmental pathways as well as the transcriptional control mechanisms underlying cell fate specification.