AI traces two-way traffic inside a growing fungus

A fungus can send particles both ways through one microscopic tube. A 2025 Nature study used KymoButler to measure this traffic in plate-grown fungi associated with carrot roots. These fungi exchange nutrients with plants while building outward-growing networks.[1]

The recording becomes a kymograph: a narrow line through a video is stacked frame after frame, turning movement into sloping traces. Their slopes encode speed and direction. KymoButler’s neural network first identifies tracks and then uses a decision model to follow them through crossings.[2]

Imagine one particle moving towards a tip and another heading back. Their traces can cross on the diagram. Joining the wrong branches would make a particle appear to reverse direction, so distinguishing a crossing from a turn is part of the measurement. The software returns the sequence of coordinates along each track.[2]

Across 28 biological samples, researchers extracted more than 100,000 trajectories. Automated tracks were manually checked; unusually fast bursts required separate visual inspection.[1]

Mean speeds stayed stable over 20 days while network length grew almost a hundredfold. Early on, tipward motion was about 30% faster than rootward motion; that bias faded below statistical significance around day ten. Network growth accompanied a change in directional balance, with little shift in typical speed.[1]

Report an error or suggest a correction