Fig. 1: Parametrising tracking fidelity.
a Sketch depicting how frame-to-frame object linking is affected by the maximum linking distance for tracking (dLinking), the distance to the nearest neighbour at the current frame (dneighb), and the intrinsic displacement length of the particle (dTrue). b Sketch of our simulation-based SPT fidelity evaluation pipeline. Ground truth trajectories generated by our simulator are deconstructed to extract particle location in each frame and fed to a tracking algorithm (LAP tracker in Trackmate) to reconstruct trajectories, mimicking tracking from experimental recordings. These reconstructed trajectories are then compared to the ground truth to compute the Fidelity Score (see Methods section “Comparing ground truth and reconstructed trajectories”). c Characteristic length of the displacement (Rayleigh) distribution for a 2D Brownian motion as a function of the particle speed (diffusion coefficient) and the acquisition framerate (see Methods section “Principles of tracking algorithms”). d Percentage of correctly recovered trajectory displacements, normalised to the maximum at the corresponding characteristic length, for a density of 0.291 spots/μm2, variable particle speeds and maximum linking distances. The red dots indicate the position of the maximum value at each characteristic length. e Evolution (mean ± SEM) of the optimal linking distance as a function of the characteristic length for different spot densities. The black dashed “theoretical distance” line represents the expected distance to capture 99.9% of the displacements for the corresponding 2D Brownian motion (see Methods section “Tracking difficulty increases with characteristic lengths”). f Evolution (mean ± SEM) of the optimal Fidelity Score as a function of the characteristic length for different spot densities. The yellow line (density of 0.291 spots/μm2) corresponds to the red dots in (d). g, i, k Maps of the optimal linking distances (left) and associated Fidelity Scores (right) for a wide range of spots densities and characteristic lengths for simulations in freespace (g), ER (i) or mitochondria (k) geometries. h, j, l Density (mean ± SEM) of correctly recovered displacements as a function of the spot density for different characteristic lengths (colours) for simulations in freespace (h), ER (j) or mitochondria (l) geometries. Simulations results are averages over 5 independent repeats in freespace and over 5 independent repeats for each of the 6 ER and 5 mitochondria geometries. Source data are provided as a Source Data file.
