the Open Band (Berkeley)

A new milestone for AI-driven science. Published in Nature, AutoDNA introduces an AI-native experimental laboratory that autonomously designs, executes, and optimizes biomolecular experiments. Powered by multi-agent AI, the system coordinates 20+ types of laboratory instruments and executes over 11,000 hardware operations. Key results: - 97.8% stepwise yield in enzymatic DNA synthesis. - 3.2× faster concurrent experimental execution. - Completed an end-to-end DNA data storage cycle with exact recovery. - Identified and experimentally validated a clinically detected KRAS G13D mutation. From natural language instructions to real-world wet-lab execution, AutoDNA brings AI one step closer to autonomous scientific experimentation. The next generation of AI scientists may not just analyze experiments. They may run the lab themselves. Paper: doi.org/10.1038/s41586-026-1… #AIforScience #AIforBiology #AutonomousLab #Nature

IntAct release 253 is out! 1,814,919 binary interactions from 24,076 manually curated publications ebi.ac.uk/intact/about#stati… (in sync with UniProt release 2026_02)