AI reveals predictable rules governing glioblastoma cell state switching
Researchers have identified a structured, predictable regulatory framework governing the cellular plasticity of glioblastoma, the deadliest adult brain tumor. By analyzing over one million cells from 12 patients, the team developed a deep-learning tool called scDORI to map how tumor cells switch between states. They discovered that low-plasticity states, such as neuronal-like identities, are actively maintained by repressors like MYT1L, which suppress alternative malignant fates. Enhancing this repressor reduces tumor growth and invasion in mice, suggesting new therapeutic strategies to lock cancer cells into less aggressive states.