AWS has now unveiled Amazon Bio Discovery, an AI platform that grants researchers direct access to a broad library of biological foundation models that can be fine-tuned for specific use cases in drug discovery. Announced at the AWS Life Sciences Symposium at the Javits Center in New York, the platform is supported by an AI agent that can select models for research goals, and evaluate candidates for synthesis and testing to enable a rapid lab-in-the-loop experimentation cycle.  

While rising AI models show promise, they require coding skills and the ability to manage computing infrastructure. Additionally, diverse models face benchmarking challenges and moving candidates from computational design to physical synthesis remains a multi-step process. Given that data live in disconnected systems, scientists must manage multiple lab partners and manually coordinate timelines and execution.

Amazon Bio Discovery addresses these challenges with three capabilities: a benchmarked library of AI models and analysis packages, an AI agent that supports experimental lab, and integrated lab partners that test top antibody candidates and route results back to the researchers. This feedback loop improves the next round of design. 

“AI agents make powerful scientific capabilities accessible to all drug researchers, not just those with computational expertise,” said Rajiv Chopra, PhD, vice president of AWS Healthcare AI and Life Sciences.  

Currently, 19 of the top 20 global pharmaceutical companies use AWS to power research workloads. Amazon Bio Discovery will bring enterprise-grade scale, privacy, and security to researchers across pharmaceutical, biotech, and academic research organizations. MSK, Bayer, the Broad Institute, and Voyager Therapeutics are among early adopters of Amazon Bio Discovery. 

Among the Amazon Bio Discovery broad catalog includes open-source and commercial models from Apheris and Boltz. Biohub and Profluent are expected to join the platform. 

Amazon Bio Discovery enables scientists to fine-tune the model by feeding prior experimental data from their organization’s lab results into the application without complex training pipelines or custom code. In-house models can also easily be deployed and hosted within Amazon Bio Discovery.  

To support model selection, an antibody benchmark dataset is available to evaluate the likelihood of a drug candidate to have favorable biological properties, such as manufacturability and stability. 

Candidates selected for experimental validation can be directly sent to Amazon Bio Discovery’s integrated network of laboratory partners, including Twist Bioscience, Ginkgo Bioworks. A-Alpha Bio is also anticipated to join the network. 

The post AWS Launches Amazon Bio Discovery Agentic AI to Accelerate Drug Development appeared first on GEN – Genetic Engineering and Biotechnology News.