Google made cheminformatics Skills available through Antigravity’s Science plugin and demonstrated a roughly one-hour workflow spanning data collection, molecular comparisons, predictive models and new structures. The language model orchestrates work while deterministic tools such as RDKit perform calculations, with connections to ChEMBL and PubChem. Guidance includes chronological or scaffold-based training and validation splits. Sparse data and differences between laboratory assays limit accuracy, so proposed molecules and synthesis routes still require human review, synthesis and experimental validation.
Back to Tech
Antigravity adds cheminformatics Skills for scientific workflows
Proposed molecules and synthesis routes are computational hypotheses, not experimentally validated drugs.