dist_mixtures

A package for fitting histograms of spatial orientations by a mixture of von Mises distributions.

License: New BSD License, see the LICENSE file.

Data

This package has been developed to analyze muscle cells directions data coming from aortic segments. The related data files can be found at [1].

[1] https://github.com/rc/dist_mixtures_data

Installation

Download the source code from [1].

[1] https://github.com/rc/dist_mixtures

The following dependencies need to be installed:

• NumPy (http://numpy.org)
• SciPy (http://scipy.org) - Git version
• Matplotlib (http://matplotlib.sourceforge.net)
• Statsmodels (http://statsmodels.sourceforge.net) - Git version

Then cd into the dist_mixtures/ directory.

Usage

Run fit_von_mises.py without arguments to see the help message::

$ ./fit_von_mises.py Usage: fit_von_mises.py [options] pattern data_dir

Fit data files with names matching a given pattern by a mixture of von Mises distributions.

Options: --version show program's version number and exit -h, --help show this help message and exit -o dirname, --output-dir=dirname output directory [default: output] -c filename, --conf=filename use configuration file with parameter sets. Ignored, if n_components option is given. -n positive_int, --n-components=positive_int number of components of the mixture [default: 2] -p kappa0,mu0,kappa1,mu1,..., --parameters=kappa0,mu0,kappa1,mu1,... initial guess of von Mises parameters for each component as a comma separated list, e.g., for two components: "1,0,1,0" corresponding to kappa0, mu0, kappa1, mu1 respectively. The location parameter mu should be given in degrees in [-90, 90[. -d pattern, --dir-pattern=pattern pattern that subdirectories should match [default: *] -m positive_int, --merge-bins=positive_int number of consecutive bins in data to merge [default: None] --plot-bins-step=positive_int step to choose bins from all bins for histogram plots [default: 4] --spread-data spread raw data using their counts instead of just repeating them -a, --area-angles compute and draw angles of two systems of fibres determined by equal histogram areas --no-neg-shift do not add 180 degrees to negative angles -s, --show show the figures

Example runs

By default, the results are stored into a directory called output. Use -o option to change that.

• Analyze histograms in *.txt files of data sets in data/<dataset name> directories, assume 2 von Mises mixture (VMM) components::

  $ ./fit_von_mises.py "*.txt" data/ -n 2

• Run analysis with parameters given in a file. This example shows how to run the analysis with varying number of VMM components::

  $ ./fit_von_mises.py "*.txt" data/ -c examples/psets/n_components.py

  See other example parameter set files in examples/psets/.