Covariance analysis of worm swimming facilitates automated phenotyping. (A) A typical image from a movie of a worm thrashing in a well of a 96-well plate. (B) The same image, after removal of the first Principal Component to eliminate most of the image background. (C) A summary of the method used to measure worm swimming rates using covariance. After acquisition with a digital camera attached to a stereomicroscope, movies are stored for subsequent analysis. (D) After subtracting the background using Principal Components Analysis, a covariance matrix is computed. (E) The number of frames separating two peaks in covariance is equal to the interval over which the worm has undergone a complete cycle of conformations, and is therefore twice the interval between thrashes. (F) A sequence of background-subtracted images corresponding to the points in (D) and (E) covering a single thrash cycle.