From particle tracking to molecular interactions

Live cell imaging often requires single particle tracking to capture the spatio-temporal details of molecular dynamics in situ. Here we present a single-particle tracking algorithm that constructs complete and unbiased trajectories from time-lapse sequences: It first links detected particles frame-by-frame, then closes gaps between initial tracks, accounting for particle appearance, disappearance, merging and splitting. Both steps are compactly formulated as a linear assignment problem, leading to a globally optimal solution. Tracking is supported by on-the-fly classification of particle motion. The tracking algorithm is flexible and versatile. We have applied it to study trans-membrane protein interactions, endocytic pit lifetime distributions and sister kinetochore motion and coupling.