Density Visulaization Dialog

This dialog presents a visualization of the max velocity density plot, along with a superimposed graph of the EM fitted gaussians. The user can either accept the threshold suggested by the program, or set the threshold manually.

• Density Plot: The plot presents the density of the max velocity withing motion segments (in black), the EM fitted gaussians (in blue), and their sum (in red). The leftmost gaussian presents the lingering episodes population, and the gaussian(s) to it's right presents the population of progression segments. The intersection between the two leftmost gaussians determines the threshold value. Reference:
  • Dan Drai, Yoav Benjamini, Ilan Golani (2000) Statistical discrimination of natural modes of motion in rat exploratory behavior. Journal of Neuroscience Methods. Volume (issue): 96 (2)

  Note that the density function is performed on the transformed data (according to the Max Velocity Transformation that was chosen). The density graph can be affected substantially by the selected transformation.

• Threshold Value: The threshold value that was chosen by the program appear in the "Transformed" and "Non-Transformed" edit boxes. The non-transformed value allows the user to use his intuition and judgement for either accepting the threshold or setting it manually. The default is accepting the threshold suggested by the program by clicking the Accept button. To set the threshold manually, you can:
  1. Type the desired value into the "Non-Transformed" edit box.
  2. Use the mouse cursor to point the desired X-coordinate on the density plot, and click the mouse. The selected X,Y coordinates apprear under the plot, on the right. Once you are satisfied with the selected X coordinate, press the Set Threshold button.
• Number of Gaussians: We recommend the use of a 2 gaussians model for mouse phenotyping. The program allows, however, to change the number of gaussians in the fitted gaussian mixture model by clicking the buttons Add one and Substract one. The two P value windows next to these buttons show the p value for the improvement in fit between the currently used gaussian mixture model and a model that contains one more or one less gaussians respectively. When the improvement in fit for adding a gaussian becomes insignificant (p value of more than 0.05), then there is no need to increase further the number of gaussians. At present, however, we recommend the use a 2 gaussians mixture model regardless of the p values shown, as it proves to work best for phenotyping various mouse strains.
• Gaussian Parameters: This window shows the parameters (mean, standard deviation and proportion coefficient) for each of the gaussians in the gaussian mixture model that is fitted to the data. For further reading on the gaussian parameters, the interested reader is referred to:
  • Dan Drai, Yoav Benjamini, Ilan Golani (2000) Statistical discrimination of natural modes of motion in rat exploratory behavior. Journal of Neuroscience Methods. Volume (issue): 96 (2)

Once you are satisfied with the selected threshold, press the Accept button. The threshold will be stored in the ".inf" file. You can also choose to either Skip this file, or Abort and return to the main dialog.