PantheonOS: An Evolvable Multi-Agent Framework for Automatic Genomics Discovery

Abstract

An AI-driven multi-agent framework integrating large language model-powered autonomous agent systems with single-cell biology for automated biological discovery. Key capabilities include agentic code evolution, batch correction, gene panel selection, and spatial genomics reconstruction, with applications spanning mouse embryogenesis, human fetal heart development, and cardiac disease modeling.

Publication
bioRxiv