A simple method for automatic tracking of actin filaments in the motility assay

A great deal of quantitative information about the actomyosin interaction can be obtained from the basic Kron and Spudich in vitro motility assay provided that care is taken to obtain consistency between experiments and that the data is examined comprehensively and not selectively. From observations of filament movement under a wide variety of conditions we have formulated the hypothesis that a large number of filaments moving over a short time period is indistinguishable from fewer filaments moving over a longer sequence of frames. This has been used to devise a simple automation of filament detection procedures. A sequence of images is digitized through a frame-grabber. If successive pairs of frames are compared the program will search for and detect the new position of every filament and show its vector on screen. Velocity is calculated and shown as a frequency histogram. The program regularly detects over 100 filaments moving in each pair of frames; usually a sequence of up to 15 pairs of frames are studied yielding 500-1000 vectors in total. The algorithm cannot deal with filaments that meet, cross or divide, however, when filaments are moving less than 2 mu m between frames this is only a small proportion of the whole. The program outputs fraction of filaments motile, mean velocity with standard deviation and density of filaments (filaments mu m(-2)). A cumulative frequency histogram gives an immediate visual indication of the performance of the population of filaments. Direct comparisons show that the data produced by automatic tracking is indistinguishable from manual tracking apart from the small apparent velocity of non-mobile filaments. The detection process takes about 5 min and requires little skill or judgement. This can lead to great increases in the rate of data analysis in motility work.