Assessed error with data extraction algorithm which extracts run velocity, run duration, tumbling angle and tumbling duration. Following steps were taken (and documented on OWW Wiki):
Used known synthetic trajectory as an input for data extraction algorithm
After obtaining motility data on the above 4 characteristics, posteriors were plotted. mu and sigma for run velocity, lambda for run and tumbling duration.
Error assessed by finding % error between parameter value for which posterior is maximised and the actual value of the parameter used to generate synthetic trajectory.
It was found that the errors are quite large. This is due to the low frame rate such that tumbles which are less than 1/(frame rate) do not cause the cell to be stationary. Only way to get around this problem with real data is to increase the frame rate at which our videos are captured.
We also tracked more cells today, tip and centre. Total number of cells tracked so far is 20, with over 5000 data points (coordinates).
Segmentation algorithm still does not provide us with satisfactory visual segmentation. Clustering software however, can find 2 centres, but we are not convinced whether this is of any significance.
This is an important issue to be resolved before we can even move onto modelling motility of bacteria.
"
