The Chan Zuckerberg Initiative unveiled GREmLN, an AI model designed to capture complex gene regulatory networks in single-cell RNA data using graph-based neural networks. Developed by Andrea Califano and colleagues at Columbia University, GREmLN predicts cellular responses to interventions like drug treatments by modeling causality and molecular logic more effectively than prior transcriptome AI models. The model advances CZI’s 'virtual cell' program and offers parameter-efficient architecture with accelerated training, potentially enabling improved AI-driven cancer immunotherapy research and precision medicine.