By Priya Dialani @priya_dialani
Published July 31, 2025 at 12:00 PM BST
Google DeepMind has announced plans to conduct its first human clinical trials using AI-developed protein therapeutics by the end of 2025, marking a significant step toward its ambitious goal of "solving biology" to combat diseases ranging from rare genetic disorders to cancer.
The UK-based AI lab, now part of Alphabet Inc., revealed that its AlphaFold system has achieved 92% accuracy in predicting protein structures, surpassing traditional methods that often require costly and time-consuming lab experiments. This breakthrough has enabled researchers to design novel proteins targeting specific disease mechanisms, including those involved in transthyretin amyloidosis—a rare condition affecting organ function.
DeepMind's collaboration with the European Bioinformatics Institute has already produced a database containing predictions for over 200 million protein structures, which is being used by pharmaceutical companies to accelerate drug discovery. The upcoming trials will focus on safety assessments of AI-designed therapeutic proteins in patients with rare diseases, with potential applications for personalized medicine and gene therapy.
"We're entering a new era where AI doesn't just observe biology but actively designs solutions," said Demis Hassabis, CEO of Google DeepMind. The team is also developing AlphaMissense, a tool that identifies disease-causing mutations in non-coding DNA regions, potentially revolutionizing cancer treatment by enabling precise targeting of tumor-specific antigens.
While ethical concerns remain regarding AI's role in human experimentation, DeepMind emphasized that all trials will adhere to strict regulatory guidelines. The company is also working on extending its AI models to analyze cellular signaling pathways and optimize immunotherapies for complex conditions like autoimmune diseases and neurodegenerative disorders.
【全文结束】
