DeepMind's Secretive AI Drug Engine Impresses Scientists
Google's DeepMind has developed a new AI-driven drug discovery engine that is reportedly wowing scientists with its predictive power for protein-drug interactions. However, the company has kept the details of the technology proprietary, preventing outside researchers from fully assessing its capabilities or methodology.
- The new engine comes from Isomorphic Labs, a company spun out of Google's DeepMind in 2021, which is led by DeepMind CEO Demis Hassabis. Isomorphic Labs focuses on applying AI to drug discovery and has established partnerships with pharmaceutical giants like Eli Lilly and Novartis worth nearly $3 billion. - This technology builds on DeepMind's previous breakthroughs, AlphaFold 2 and AlphaFold 3. AlphaFold 2 successfully predicted the 3D structures of over 200 million proteins, a task that previously could take years for a single protein. - The latest model, called the IsoDDE, reportedly doubles the performance of its predecessor, AlphaFold 3, and can more accurately predict "binding affinity"—how strongly a potential drug molecule will attach to a target protein—than traditional, time-consuming physics-based methods. - Unlike the AlphaFold models, which were described in scientific papers and made accessible to researchers, the new drug engine is proprietary, meaning its underlying methods have not been publicly disclosed. - The traditional drug development process is notoriously slow and inefficient, often taking over ten years with a failure rate of approximately 90% for drug candidates entering clinical trials. AI aims to shorten this timeline from years to months by rapidly screening compounds and predicting their effects digitally. - This work highlights the role of computational biologists and bioinformaticians, who combine biology, computer science, and statistics to analyze massive datasets like protein structures and genetic sequences. They build the AI models and digital tools used in the early, pre-clinical stages of drug discovery. - Once an AI-designed drug candidate is identified, it moves into clinical research, a more patient-facing field. Professionals in this area, such as Clinical Research Associates, design and manage the clinical trials that test the safety and efficacy of new treatments in humans, ensuring they meet regulatory standards. - Some experts predict that as AI accelerates the discovery of new drug molecules, the demand for clinical research professionals will boom after 2030 to manage the increased volume of drugs entering human trials.
