Amazon Web Services (AWS) announced the launch of its AI bio tool, Amazon Bio Discovery, to accelerate the early-stage process of drug discovery in the pharmaceutical industry. The application aims to help scientists design and test novel drugs quickly and securely.
Amazon Bio Discovery enables scientists to run complex computational workflows through more than 40 AI-specialized foundational models, trained on various biological datasets. These models generate and evaluate potential drug molecules, alongside AI agents that help scientists select models, optimize inputs, and evaluate candidates according to their research.
Researchers can send the selected candidate list to integrated lab partners for synthesis and testing, and the results are sent back to the application for analysis and model refinement, creating what AWS calls the lab-in-the-loop.
With the increased rise of AI biological models, each with different characteristics, computational biologists responsible for operationalizing these models experience bottlenecks. Meanwhile, bench scientists with deep expertise encounter slow processes for research or experimenting due to a lack of direct access to computational tools. Amazon Bio Discovery aims to solve this problem by providing a platform that brings computational design and wet-lab validation.
The Memorial Sloan Kettering Cancer Center, which recently partnered with Amazon Bio Discovery, states that they designed nearly 300,000 new antibody molecules and sent the top 100,000 candidates for testing, enhancing the workflow timeframe from one year to just weeks.
Rajiv Chopra, vice president of healthcare AI and life sciences at AWS, reportedly said that the service is intended to increase support for scientists and contract research organizations, rather than replace them. Additionally, Tycho Peterson, Jefferies analyst, said that AI fears of reducing the need for instruments in drug research are inflated, as there is scope for increased investment and return as research programs escalate.
AWS mentioned that among the early adopters of Amazon Bio Discovery are Bayer, the Broad Institute, and Voyager Therapeutics, alongside the endorsement of its cloud services by 19 of the top 20 global pharmaceutical companies.
Additionally, AWS and Gray Lab at Johns Hopkins Engineering launched the Antibody Developability Benchmark dataset, one of the most diverse and large-scale databases for AI-informed antibody design. This new database is part of Amazon Bio Discovery, and additional benchmarks will be added over time.