Neighborhood-informed positional information for precise cell identity specification

During development, cells reliably establish their identities, a process that is enabled in part by positional information encoded in gene expression patterns. Previous works showed that cells in Drosophila embryos can utilize this information to decode their position along the anterior-posterior axis with a 1% embryo-length positional precision. However, this precision is insufficient to uniquely determine position, leading to a positional information gap. Here, we propose a neighborhood-informed information-theoretic framework which allows to quantitatively estimate the amount of information regarding position which exists in the microenvironment of each cell. We formulate how much additional information exists in neighboring cells as a function of spatial variation in gene expression. We show that the additional information encoded by local neighborhoods is sufficient to uniquely specify cell identities, closing the information gap on average across major patterning axes in Drosophila embryos, gastruloids, and the developing neural tube. Furthermore, neighborhood-informed decoders predict cell positions and downstream gene expression patterns more accurately than cell-independent decoders, resulting in lower decoding variability, which is maintained in mutant Drosophila embryos. Our results provide a basis for the analysis of cellular decision-making in the context of their microenvironments.