A team from Warwick and Monash identifies the molecular mechanism that lets bacteria assemble an à la carte family of depsipeptide anticancer agents — a foundation for engineering variants, still far from the clinic.
Read the articleA 3D model whose encoder is trained without labels on 5.24 million routine CT and MRI volumes reaches 92.68% AUROC across 82 CT diagnoses and 92.49% across 74 MRI diagnoses. With the archive held identical, it is the training objective — latent-space prediction — that opens the gap.

On 13 August 1856, Henry Bessemer presented to the British Association for the Advancement of Science a process that converts 5 tonnes of pig iron into steel in 20 minutes by blowing air through it — cutting the price of steel by a factor of 6 to 8, from ~£40-60 to £7 per tonne.
In May 2024, DeepMind publishes AlphaFold 3. Beyond isolated proteins, the model predicts the structures of protein–DNA, protein–RNA, and protein–ligand complexes with accuracy surpassing all existing methods — opening the door to large-scale computational drug design.