AWS launches Bio Discovery

Drug discovery often starts on computers before it reaches a lab bench, with researchers testing how proteins and molecules might fit together like puzzle pieces. On April 14, Amazon Web Services launched Amazon Bio Discovery to let scientists run more of that early work without writing code. (reuters.com) Amazon Web Services said the application gives researchers access to more than 40 biological foundation models, which are artificial intelligence systems trained on large biology datasets to predict properties such as binding and stability. The company said users can also bring in their own models or licensed third-party models. (aws.amazon.com) The software is built around step-by-step workflows for tasks such as molecule design, lead selection, and optimization, instead of requiring scientists to assemble code pipelines themselves. Amazon said agentic assistants help choose models, tune settings, and rank candidates for follow-up experiments. (aboutamazon.com) Early-stage drug discovery is the phase when companies search through huge numbers of possible molecules to find a few worth testing in the lab. Amazon said Bio Discovery links that computer screening to contract research organization partners so wet-lab results can feed back into the next design cycle. (aws.amazon.com) Amazon presented the launch alongside its Life Sciences Symposium in New York on April 14, where the company framed agentic artificial intelligence as a new tool for drug research and development. The product extends Amazon Web Services deeper into life sciences software, an area where cloud providers have been selling infrastructure, data tools, and machine learning services for years. (aws.amazon.com) Drugmakers and technology companies have spent the past several years trying to use artificial intelligence to cut the time and cost of finding new medicines, but many projects still require specialist computational teams. Reuters reported that Amazon is pitching Bio Discovery as a way to lower that technical barrier for biologists and chemists. (reuters.com) Amazon said the system has already been used in a scientific validation study that generated more than 300,000 computationally designed nanobody candidates against a novel cancer antigen. The company used that result to argue the platform can help teams evaluate far larger candidate sets before moving into physical experiments. (awsstatic.com) The launch also drops Amazon into a crowded field of artificial intelligence drug-discovery companies and cloud rivals chasing pharmaceutical research budgets. Axios said the release sharpens competition around promises of faster and cheaper early-stage development, while Amazon is betting that easier software will pull more scientists onto its cloud. (axios.com)