Glowing spots on a dark background: that's what single-molecule images from MINFLUX look like. This microscope can reveal details smaller than five nanometers: roughly ten thousand times thinner than a human hair. But the picture isn't ready-made; it's more like a scatter of lights on a Christmas string where some bulbs have burned out and the wires are tangled.
Making sense of such a scatter is tricky: molecules in a cell aren't arranged in a neat line, labels don't all glow at once, and the instrument itself sometimes errs. Previously, researchers had to manually tweak parameters and know in advance what shape to look for. The new approach works differently: it groups the blinks into clusters on its own, estimates uncertainty, and finds recurring patterns without any hints.
The method was tested on both simulated and real MINFLUX data. The accuracy of grouping blinks exceeded 0.75 under all conditions, and structure recognition reached 0.9 when label density was good. This means scientists can study unknown molecular machines without knowing their architecture beforehand.
In this way, a scatter of lights turns into a clear diagram, as if the outline of a Christmas tree emerged by itself from a tangled string of lights.
MINFLUX uses a minimum of photons to avoid damaging living cells, yet it still sees details 100 times finer than a conventional optical microscope.
Science fiction often features scanners that instantly produce a 3D model of an unknown object: this method takes a step toward such technology for the molecular world.