Evolution of Dynamic Models for Embryonic Patterning and Their Geometric Representation
summary
Uses an evolutionary algorithm to mutate network topology and parameters of pseudo‑biochemical nodes, evaluating fitness as mutual information fitness between selector‑gene identity and position; cell‑autonomous, multistable networks emerge that read a static morphogen or a sliding morphogen timer and produce discrete territories (example: evolved five territories, fitness near −log7 in one run). Models reproduce Hox phenomenology — temporal collinearity and posterior prevalence — and suggest a single timer gene can convert short‑ to long‑germ patterning; results are robust across many simulations but not universal.
AI-written summary · about