{"data":{"id":904,"lang":"en","title":"Nvidia Unveils Proteina-Complexa AI Model That Designed 1 Million Drug Candidates Validated in Lab","summary":"Nvidia's new open-source AI model designed one million protein drug candidates, with lab tests confirming binders for 68% of targets.","category":"technology","category_label":"Technology","published_at":"2026-03-19T13:02:21Z","updated_at":"2026-03-19T13:02:21Z","url":"https://techtools.cz/tools/news/article/904-nvidia-unveils-proteina-complexa-ai-model-that-designed-1-million-drug","api_url":"https://techtools.cz/news-api/v1/articles/904?lang=en","markdown_url":"https://techtools.cz/news-api/v1/articles/904.md?lang=en","image_url":"https://iprsoftwaremedia.com/219/files/202603/69b5b8393d6332de6974de0b_NVIDIA-open-model-families-agentic-physical-healthcare-ai/NVIDIA-open-model-families-agentic-physical-healthcare-ai_mid.png","word_count":127,"reading_time_minutes":1,"sources":[{"url":"https://nvidianews.nvidia.com/news/nvidia-expands-open-model-families-to-power-the-next-wave-of-agentic-physical-and-healthcare-ai","domain":"nvidianews.nvidia.com"},{"url":"https://www.pharmiweb.com/press-release/2026-03-16/manifold-bio-demonstrates-million-scale-experimental-validation-of-ai-driven-protein-binder-design-w","domain":"pharmiweb.com"}],"translation":{"lang":"cs","id":934,"title":"Nvidia představila AI model Proteina-Complexa, který navrhl milion kandidátů na léky ověřených v laboratoři","url":"https://techtools.cz/tools/news/clanek/934-nvidia-predstavila-ai-model-proteina-complexa-ktery-navrhl-milion-kandidatu-na","api_url":"https://techtools.cz/news-api/v1/articles/934?lang=cs"},"content":"Nvidia announced Proteina-Complexa at GTC 2026, an open-source generative AI model that designs protein binders for drug discovery. In a landmark validation, partner Manifold Bio tested one million AI-generated binder designs against 127 protein targets in a single experiment, measuring over 100 million protein-protein interactions and identifying specific binders for 68% of targets tested.\n\nThe model uses flow matching to jointly model backbone geometry, side-chain conformations, and amino acid sequences, enabling rapid design of therapeutic proteins. Pharma giants Novo Nordisk and Viva Biotech are already using the platform to accelerate their drug pipelines.\n\nAlongside Proteina-Complexa, Nvidia collaborated with Google DeepMind, EMBL's European Bioinformatics Institute, and Seoul National University to add 1.7 million high-confidence protein complex predictions to the AlphaFold database, calculated from roughly 30 million total predictions."}}