Notes
Science in plain language, and anything else worth writing down.
- A pink dye, a beam of light, and molecular trees that carry it
Rose bengal can kill cancer cells when lit, but struggles to get inside them. Our simulations show how branched carrier molecules grab hold of it.
- Can a language model tell how long a drug stays bound?
Predicting how long a drug stays on its target usually calls for a 3D structure. We asked whether models that read sequences and chemical strings can capture some of it anyway.
- How many strands make a microtubule, and why it matters
Microtubules can be built from different numbers of strands. Our simulations suggest the choice trades structural support against how well mechanical signals travel.
- How taxol changes the mechanics of microtubules
One of the most used cancer drugs works on the cell’s skeleton. We followed its effect from the binding pocket to the way the whole tube flexes.
- Teaching a computer to predict bitterness
How we built machine-learning models that predict which bitter receptor a molecule will activate, and that show which parts of the molecule drove the answer.
- Turning a cell’s scaffolding into a capsule that opens on cue
How a team in Japan rebuilt tubulin, the protein of microtubules, into hollow capsules that open in response to GTP, and what our simulations added.
- What anaesthetics do to a cell membrane
Volatile anaesthetics are chemically very different, yet they act alike. Our simulations show what they do to the membrane around the proteins they target.