ttdtrang / pygenenet

Python Implementation of GeneNet algorithm

PyGeneNet

This package implements GeneNet algorithm for learning causal genetic network from time series data. The original implementation is described here

Barker N. A., Myers C. J., and Kuwahara H., “Learning Genetic Regulatory Network Connectivity from Time Series Data,” IEEE/ACM Transactions on Computational Biology and Bioinformatics, vol. 8, no. 1, pp. 152–165, Jan. 2011.

Installation

The program requires python 2.7 and packages below

* numpy 1.8.2
* pandas 0.15.2
* graphviz 0.4.3
* matplotlib (only required for timing in ``examples``)

To use this program, please have python installed on your system, and run:

python setup.py install

The package will be installed in Python site-packages directory by default and will be available under the name pygenenet.

To check your installation:

python
>>> import pygenenet

should not return an error

To remove the package, run (with appropriate permission)

pip uninstall pygenenet

Or manually remove the files in the location installed above. The exact locations of each files can be obtained by running installation again with the --record option

python setup.py install --record files.txt

Location of files will be written to files.txt.

Usage

Example use can be found in pygenenet/examples

Data files:

net3_ssa_10 : synthetic data from stochastic simulation of a 3-species network, for 2000s, trajectories written every 10s net3_ssa_100 : synthetic data from stochastic simulation of a 3-species network, for 2000s, trajectories written every 100s net4_ssa_10 : synthetic data from stochastic simulation of a 4-species network, for 2000s, trajectories written every 10s net4_ssa_100 : synthetic data from stochastic simulation of a 4-species network, for 2000s, trajectories written every 100s

Example scripts:

1. demo_net3.py to learn the network from net3_ssa_10. The result should look like this

      Learned causal network in 23.927177906s
               CI  LacI  TetR
         CI     0     0    -1
         LacI  -1     0     0
         TetR   0    -1     0
   

2. demo_net4.py to learn the network from net4_ssa_10. The result should look like this

      Learned causal network in 23.927177906s
            CI  GFP  LacI  TetR
      CI     0    0    -1     0
      GFP    0    0     1     0
      LacI   0    0    -1     0
      TetR   0    0    -1     0
   

3. performance_net3.py and performance_net4.py will time the learning algorithm with various input subsets and input data frequencies and report a plot. These scripts require matplotlib.