Biology Is Moving from PDF to API

Biology Is Moving from PDF to API

The main thing holding progress back today is not a shortage of ideas, but a floppy-disk-era way of packaging knowledge. We are building extraordinarily sophisticated AI models to predict protein structure on top of datasets that were assembled over decades in proprietary and often machine-unreadable archives. Jed McCaleb (Astera Institute) decided to change that by launching Radial, an infrastructure division with a budget of 500 million dollars.

Radial’s mission under Becky Pferdehirt (ex-a16z) is not to find a “magic pill,” but to rebuild the operating system of science. Biology is stuck in the format of static PDF papers and closed datasets that modern algorithms cannot digest. Radial is going after the “unsexy” but critical bottlenecks: data, publishing, and collaboration.

Its first flagship is the Diffuse project. If the existing PDB (Protein Data Bank) is an album of a billion photographs, Diffuse is meant to become a video library. A protein structure is not a frozen sculpture, but continuous motion. Without understanding that dynamic behavior, we cannot design drugs effectively. Radial will build the experimental and computational stack needed to make that motion visible and usable for training neural networks.

The second direction, The Stacks, is an attempt to kill the traditional scientific journal. The platform moves away from giant “paper” articles toward nano-publications: short, verifiable results released continuously in a machine-readable format. That turns science from slow postal correspondence into a live, distributed software-development process, where errors are caught in real time instead of years after publication.

The third vector is Open ADMET. Today, data on how molecules behave in the body is the “gold reserve” of pharma giants, securely locked behind paywalls. Radial is building open, high-quality datasets that will let the entire industry train models on real data instead of reading tea leaves.

For us, this is a sign that the longevity field is maturing. We can no longer wait for flashes of insight from lone geniuses in garages. We need an industrial base: Github for atoms and a PDB for living systems in motion. When science moves from PDF to API, the pace of discovery will change not by percentages, but by orders of magnitude. Radial is an attempt to build the foundation on which the biology of the future will stand.