Figure 1.
WormSizer: an open source application for detecting nematode size and shape.
A. The standard workflow of the application. The only manual step (the optional review) is depicted as a trapezoid. B. A system diagram depicting how WormSizer interacts with Fiji. WormSizer is written in Java as a Fiji plugin. Fiji plugins written in Java can utilize other Fiji plugins as well as any 3rd party Java library. C. The output of WormSizer. Each worm in an input image is identified separately from others. Its contour is highlighted in yellow, its skeleton in blue, and the sampled radii in cyan. This image is shown to users and may be manually passed or failed. D. The volume calculation of WormSizer.
Figure 2.
A. The window in which users input the image processing options. Note, there is an option to real-time preview changes in settings. B. The review of the image processing results. The user may optionally pass or fail each worm identified in an image. Each worm is shown sequentially (with the option to go backwards or forwards) and may be passed or failed using a keyboard or mouse.
Figure 3.
Volume measurement is robust to varying interval sizes.
WormSizer was used to segment 10 images of different nematodes (denoted by individual lines in the figure). The interval size (x-axis) was then varied and the resulting calculated volume reported. The average length of the worms was 396 pixels, so an interval size of 40 is approximately 10% of a worm's length. The mean CV of volume across worms across interval sizes was 2.3%.
Table 1.
Repeated measurements of individual worms.
Table 2.
The number of samples and biological replicates per strain.
Figure 4.
Protocol for imaging staged populations of worms.
Each strain is bleached and hatched in the absence of food so it enters L1 arrest (0 hr recovery). Arrested L1 stage larvae are recovered on plates with E. coli OP50 as food to initiate post-embryonic development and imaged at subsequent time points. Worms are washed on to clean plates for imaging.
Figure 5.
Length measured over time reveals significant differences in growth rate between mutants and wild-type.
A boxplot is shown with each strain plotted next to its control over time. The middle hinge of each box is the median, and the lower and upper hinges mark the lower and upper quartiles respectively. The lower whisker is the lower 25% quartile minus 1.5 times the interquartile range (the difference between the upper and lower quartiles). The upper whisker is the upper quartile plus 1.5 times the interquartile range. Outliers that appear outside the whiskers are marked as dots.
Table 3.
Statistical significance of size measurements.
Figure 6.
Differences in size upon hatching and after 48 hr growth.
A boxplot shows the volume of each strain at 0 and 48 hours. The middle hinge of each box is the median, and the lower and upper hinges mark the lower and upper quartiles respectively. The lower whisker is the lower 25% quartile minus 1.5 times the interquartile range (the difference between the upper and lower quartiles). The upper whisker is the upper quartile plus 1.5 times the interquartile range. Outliers that appear outside the whiskers are marked as dots.
Figure 7.
Analyzing length and width over time approximates nematode morphology.
Length and middle width of individual animals is plotted for each strain and its control, with time points indicated by color. A linear regression is included for each strain, and the slope of this regression provides an indication of shape.
Table 4.
Statistical significance of shape measurements.
Figure 8.
WormSizer can detect differences in size with relatively few samples.
Results of a power analysis (t-test, two-sample, two-sided alternative, Type I error probability at 0.01, Type II error probability at 0.2) are plotted for each time point. Less than a hundred individuals are required to detect a 15% difference in volume between populations, and less than three hundred are required for a 10% difference.
Figure 9.
WormSizer can be used on mixed stage populations.
Length and width of individual animals are plotted for a mixed staged population (purple) and a staged population over time (green). Shape (length versus width) is not affected by culture method, as revealed by a pair of linear regressions with similar slope (Table 4).