Quantification of myosin motion.

All tracking analysis was performed in python. The code and example data can be found at https://github.com/OakesLab. Briefly, images were first filtered with a 15 pixel square Laplacian of Gaussian filter with a standard deviation of 2, to emphasize the myosin filaments. Filaments were then tracked using the trackpy software package (https://github.com/soft-matter/trackpy) with the following relevant parameters: feature_size = 11 pixels, memory = 2, and separation = 3. As trackpy doesn’t use sub-pixel localization the resulting tracks were filtered using a running window average over ~2.5 seconds (translating to 5–8 frames depending on the imaging frequency). The resulting tracks were further filtered to only consider tracks with a path length of at least 900 nm (e.g. the length of ~ 3 myosin filaments) to ensure that we were only considering persistent motion. While tracks shorter than this certainly occurred, we restricted measurements to this range to ensure we could have confidence in the measurements of the particle velocities.

To determine the direction of the flow, we first defined the image intensity center of mass. A vector drawn from the center of the image to the center of mass determined the mean direction of retrograde flow. The direction of the myosin movement was defined as the vector drawn from the first position in the smoothed track to the last. By calculating the dot product of these two vectors we were able to define an angle, θ, relative to the mean direction of flow. We considered any angle θ < π/3 to be retrograde flow, while any angle θ > 2π/3 to be anterograde flow (essentially creating 120 degree cone in each direction). Tracks with angles between these two directions were not considered in our analysis, as they were predominantly moving parallel to the edge of the cell.

Mean values for both retrograde and anterograde flow rates were determined by fitting a gaussian to the histogram of the pooled data. The data is reported as the mean of the gaussian fit ± the standard deviation of the gaussian curve.