Current angiogenesis research using genetic loss-of function approaches in mouse and zebrafish models provides for a growing number of genes implicated in blood vessel formation. Mechanistic insight often remains unsatisfactory as phenotypes fall into common categories. Our recent cell competition models combined with computational analysis have transformed the way we can study mechanisms in vascular patterning.I propose to systematically exploit mosaic analysis in vivo to deconstruct vascular pattern formation in development and disease. Single differentially labelled cells in mouse and zebrafish will be analysed to establish the first catalogue of endothelial cell shapes in reference to the position and developmental phase of the plexus. Dynamic imaging will describe functional shape transitions. Advanced computational analysis of 3D-segmented cell shapes using an optimized set of shape descriptors will be performed to find natural clusters via an unsupervised expectation maximization algorithm. Cross-correlation of shape clusters with gene expression, signalling, oxygen, polarity and cytoskeleton markers will be used to understand how cell shape relates to signalling and local environment. Pharmacological and clonal genetic gain and loss-of-function will be used to analyse dynamic regulation of cell shape leading to altered vascular patterning. Mosaic analysis in retinopathy and tumour models will enable unprecedented resolution to study vascular malformation. Sequential time-lapse imaging in combination with clinical imaging modalities will be used to bridge the gap between experimental and clinical tumour vasculature imaging, enabling us to ask how local environmental changes in the tumour affect blood vessel patterning.Successful completion of this work will establish the cellular and molecular principles governing vessel remodelling and provide a new conceptual framework and methodology for the analysis of pathological vascular patterning.