AI Nano Bio Agents - Small Language Model Agents for Genomics

Explore a 14-minute video presentation introducing the innovative Nano Bio-Agent (NBA) framework developed by researchers at ETH Zürich for genomics applications. Discover how this cutting-edge framework incorporates task decomposition, tool orchestration, and API access into established systems like NCBI and AlphaGenome through the GeneTuring benchmark. Learn about the potential of Small Language Models (SLMs) for handling complex genomics tasks and examine the practical limitations of model size reduction for Vision Language Models (VLMs) and Large Language Models (LLMs) in scientific applications. Investigate the feasibility of implementing local LLMs for specialized genomics work and understand the technical considerations involved in deploying smaller AI agents for biotechnology research. Gain insights from the work of George Hong from ETH Zürich and Daniel Trejo Banos from the Swiss Data Science Centre as they demonstrate how nano-scale bio-agents can be effectively utilized in genomics research and AI-driven scientific discovery.