Agentic cheminformatics with Google Antigravity

| Source: Google DeepMind Blog

Tags: Google DeepMind, Google Antigravity, Cheminformatics Skills, RDKit, drug discovery, matched molecular pair analysis

Google DeepMind added Cheminformatics Skills to Google Antigravity, letting an agent search compounds, run matched molecular pair analysis and propose new macrocycles with synthetic routes in about an hour.

Details

Google DeepMind introduced a suite of Cheminformatics Skills for Google Antigravity, turning it into a workbench for computational and bench chemists with access to local tooling. In a demo built on real lung cancer treatments, an agent searched thousands of compounds, performed matched molecular pair analysis and proposed novel rigidified macrocycles with synthetic routes in roughly an hour, versus a workflow that usually takes days per iteration. The design splits work: the language model orchestrates workflows, navigates databases and reasons scientifically, while deterministic chemistry tools such as RDKit perform the molecular calculations. The aim is to handle the many representations of a molecule, from IUPAC names to Markush structures and SDF files, where one wrong atom changes the compound.