Finding dynamical landscapes for cell fate decisions from single cell data

At the heart of embryonic development is the remarkable transformation of populations of naïve progenitor cells into complex, organized arrays of functionally diverse cell types that form the tissues fundamental to multicellular life. Substantial progress has been made in identifying molecular factors governing cell fate decisions, but the underlying principles that determine how cells make decisions are less well understood. Moreover, while data from new single cell technologies offer exciting opportunities, methods to extract information from such data are at an early stage. New approaches to uncover structure and mechanism are needed. Existing methods for predicting cell fate control infer models directly from data using clustering or tree-based methods. These provide no mechanistic insight beyond prediction and visualisation. Our approach is different. Cell fate decisions are governed by gene regulatory networks, hence the underlying biological system is a stochastic dynamical system. We propose combining dynamical systems theory and novel statistical techniques to single cell data to develop an experimental and theoretical framework that will allow quantitative, predictive and interpretable models of decision-making systems. Our overarching objective is to establish an integrated experimental-computational workflow to construct, test, and refine models of cell fate decision making. We will apply it to the complex set of transitions characterising the developing vertebrate neural tube. In Aim 1 we will develop affordable and robust experimental methods for single-cell targeted transcriptome assays that provide dynamical information. Unlike conventional whole-transcriptome single-cell sequencing, which is prohibitively expensive and generates excessive uninformative data, this will target hundreds of selected transcripts at a scale of ~million cells. This throughput is necessary to generate the data for constructing and testing the models we will develop. In Aim 2 we will develop the tools and theory to construct and parameterise dynamical models of cell fate decisions. Recent theoretical insights indicate that there are a limited set of qualitatively distinct model topologies for cell fate decisions. This realization provides a guiding framework from which we will develop new methods for embedding and modelling high-dimensional single cell transcriptome data in relevant dynamical models by using advanced statistical techniques. In Aim 3, we will combine experimental and theoretical approaches and apply these to cell fate decision making in the vertebrate neural tube. This is a non-trivial test case for which we have significant experience and provides a real-world system in which we can develop novel insight into cell fate decision making. The proposed project will be an interdisciplinary collaboration combining state-of-the-art single cell molecular biology and dynamical systems methods to produce new analytical tools that will reveal principles of cell fate decision making and the gene regulatory networks that underpin them. The project will advance the frontiers of bioscience relevant to BBSRC’s theme of Understanding the Rules of Life and contribute to the goal of Engineering Biology by developing experimental and analytical methods for designing, building and testing engineered cells and tissues. The work will establish a novel conceptual framework for understanding cell fate decision making and the operation of gene regulatory networks. This is fundamental to developmental and stem cell biology and is crucial to understanding disease processes and to the developing field of synthetic tissue biology. New tools for the analysis of single cell data, relevant dynamical systems and the identification of regulatory genes and networks will be produced and made publicly available.