The problem, plainly

A protein is a chain of building blocks that folds itself into a shape, and the shape is everything: it decides what the protein does and what medicine can attach to it. Predicting that shape from the chain alone was, for fifty years, one of biology's hardest open questions.

Where computing comes in

Simulating how a chain folds means trying astronomical numbers of possibilities against the laws of physics. That is not bench work; that is compute. Months of laboratory guesswork become weeks of simulation, and dead ends are found before anyone pipettes anything.

What our cluster did

Our dedicated research cluster ran folding and binding simulations for partner researchers: long, unglamorous batches of work that need machines which never get bored. The findings, methods and limitations are laid out in the full paper.

Where it helped medicine

Better shape predictions have already narrowed candidate molecules for further study and cut wasted synthesis in the projects we supported. The honest caveat: prediction shortens the search; the laboratory still has the last word.