Parthan Kasarapu
HOME RESEARCH PUBLICATIONS TEACHING SOFTWARE
★ gmm-search: C++ software for MML-based multivariate Gaussian mixture modelling

Cite: P. Kasarapu, L. Allison, Minimum message length estimation of mixtures of multivariate Gaussian and von Mises-Fisher distributions, Machine Learning, 2015, Vol. 100, No. 2-3, Pages 333-378.

Instructions on downloading and executing the program.
  1. Download the executable.

  2. Run the executable as follows:
    Usage: ./gmm-search --mixture --infer_components [options]
    Allowed options:
    - -help   produce help message
    - -profile arg   data file containing N rows and D columns
    (values separated by white space)
    - -log arg   output file
    - -mt arg   optional multithreading (default 1 core)
    - -improvement arg   optional threshold to terminate the search (default 0.1%)
    (Note the two hyphens at the beginning of each option.)

  3. An example:
    ./gmm-search   --mixture   --infer_components   --profile iris.dat  --log output.log  --mt 2   --improvement 0.001

  4. Output
    The output.log contains the information about the operations that are performed as part of the search method to determine the optimal number of mixture components. Each iteration has a Parent mixture, and the last such mixture in the output file is the final answer.

★ piecewise-nonlinear-fit: C++ software for Bézier segmentation

Cite: P. Kasarapu, M. G. de la Banda, A. S. Konagurthu, On representing protein folding patterns using non-linear parametric curves, IEEE/ACM Transactions on Computational Biology and Bioinformatics, 2014, Vol. 11, No. 6, Pages 1218-1228.

Instructions on downloading and executing the program.
  1. Download the archive containing the program. After uncompressing, you should have:
    • The executable piecewise-nonlinear-fit
    • A test folder containing sample PDB files
    • An empty folder named output in which the results will be stored.

  2. Run the executable as follows:
    Usage: ./piecewise-nonlinear-fit [options]
    Allowed options:
    - -help   produce help message
    - -pdbfile arg   path to pdb file
    - -maxdegree arg   maximum degree of the Bezier curve [1, 2 or 3]
    - -verbose   detailed output
    (Note the two hyphens at the beginning of each option.)

  3. An example:
    ./piecewise-nonlinear-fit    --pdbfile  test/pdb2wyq.ent.gz    --maxdegree  3
    This runs the program on the PDB file and generates the segmentation.

  4. Output
    After the program is executed, cd to the output directory.
    1. pdb2wyq-nonlinear-fit.log file contains the segmentation results.
    2. pdb2wyq.pml file can be opened using Pymol to visualize the cartoon representation of the protein along with its segmentation.
      The following command can be used:   pymol pdb2wyq.pml



Note: The programs are expected to work in a Linux environment (kernel version 2.6.18 and later).