Explainable Spatio-Temporal Graph Evolution for Developmental Neuroscience

This project studies the development of a tiny worm called Caenorhabditis elegans, focusing on how its cells divide, grow, and interact during its early embryonic stages. To map out and understand these complex interactions, the project upgrades the OpenWorm DevoGraph library using an advanced deep learning tool called the Explainable Spatio-Temporal Graph Evolution Learning (ESTGEL) model. This model tracks two critical dimensions of growth simultaneously: the 3D spatial arrangement of cells (how they physically touch and communicate in space) and the directed cell lineage (the family tree detailing how cells evolve to form different tissues). By combining these spatial and lineage pathways with an "edge attention" mechanism, the tool highlights the most critical cell-to-cell connections at various stages of growth. Ultimately, the project aims to get a complete, computationally explainable picture of the worm's development, from a single cell to a fully formed organism, and to successfully classify between healthy and mutant embryos. This work will help us learn more about basic life processes, track how early cell interactions influence neural wiring, and improve fields reliant on understanding cellular growth. The project is supported by the OpenWorm community, an international organization dedicated to using advanced dynamic graph neural networks to model and study the brain and nervous system development of the first virtual organism in a computer.