A cell atlas shows what a cell is, not what it does next. @GSK is paying @RelationRx up to $110 million to record how human cells respond, time point by time point, after a gene is changed or a drug is added.

Constitutive AMPK activity preserved late-life division in some yeast cells but inhibited fatty-acid synthesis in others. Shielding Acc1 changed the outcome.

A cardiac graft needs more than new muscle cells. Spatial maps and follow-up experiments point to Midkine as one signal recruiting host vessels.

Raygun redesigned known fluorescent proteins at new lengths. Six of eight selected shortened variants still glowed in human cells.

NSF is putting company-based research inside the dissertation: four years, two advisers and at least one year of work at an industry site.

DeepMind moved most original AlphaFold authors into broader programs. The model and database remain, but the team is no longer one narrow mission unit.

N-acetylcysteine lowered the oxidative damage and pushed death back, while the measured markers of cell-state change stayed exactly where they were.

Models fine-tuned on single cells sorted them into age groups across four species. Masking the ribosomal genes cost 22 to 28 points of accuracy; masking a random set of the same size cost 0.3. Trained on one species and tested on another, though, the models did poorly.

The contract research organization @ICONplc says it will use an AI system for site selection, enrollment risk and protocol modelling: the decisions taken before a trial's first patient. It is an operational plan, not evidence that trials have run faster.

A campaign has set a fundraising target to endow a research fund for women's aging, with each million pegged to a mile of a mountain challenge. It is a target, not money raised, and no result follows from it yet.