Genedata, a Danaher company, said Thursday its research software can connect candidates from NVIDIA’s RFD4-Proteina model to laboratory test records, as it seeks to bring AI-generated protein research into drug-discovery workflows.
Researchers can enter target traits and desired properties through Genedata’s AI co-scientist. NVIDIA models generate and assess binder candidates, while Genedata links each experiment’s results to the design tested so researchers can use that information in later rounds.
Genedata said research organizations have struggled to use generative models in production settings. “Researchers need a way to connect molecular design, experimental execution and scientific data into a continuous learning process,” Genedata President Othmar Pfannes, Ph.D., said. The company did not identify customers using the integration or disclose a broad rollout date, pricing or measured performance.
NVIDIA and the University of Washington Medicine’s Institute for Protein Design developed RFD4-Proteina as an open model that generates protein sequences and structures. NVIDIA’s model card lists antibody and small-molecule binder design among its supported tasks. It also describes reward-guided search, which scores and refines candidates against selected objectives.
NVIDIA’s model card lists Tuesday, Oct. 20, 2026, as the release date for its build.nvidia.com and NGC offerings; NVIDIA and UW Medicine announced the model publicly Wednesday, Oct. 21. The listed offering date precedes the public announcement by one day.
Danaher is also pursuing a separate AI-powered autonomous lab at Abcam. The company said Oct. 7 that the lab is expected to operate at scale in early 2027, with Genedata among the operating companies contributing technology. Danaher said the lab may eventually complement traditional immunization-based antibody discovery.
