A framework for building multi-tissue atlas of zebrafish embryo

This paper presents a new framework for building a representative atlas of multiple tissue types for zebrafish embryos. In our framework, multiple normalization strategies were considered and evaluated. Moreover, to keep the resulting atlas representative, outlier samples were discovered and removed automatically. Our quality measure for multiple tissue channels was used for selecting the best resulting atlas and for driving the normalization process (to reduce the number of experiments). We applied our framework to computing a 2D atlas of five tissue types (vasculature, sensory neuronal precursors, somites, motor neurons, and yolk) from a population of 40 zebrafish embryos.