Experiments testing ability of DSD to capture functional relationships. Required (additional) modules include: NumPy, NetworkX, and matplotlib.
This repository's design revolves around 3 modules:
dsd.py- contains basic functions for manipulating graphs and for obtaining various distance matrices -- you shouldn't need to call functions from this module directly.
expt.py- serves as a wrapper for the dsd module, adding functions to compute matrices for functional overlap/similarity, as well as the option to specify numpy binary files for previously-computed matrices.
plotting.py- contains routines that makes use of functions from the expt module, as well as numpy and matplotlib, to generate plots. Each function generates a single plot -- refer to each docstring for more information.
as well as semsimcalc.py (from semsimcalc), which is used as a submodule for obtaining Resnik similarity scores.
Scripts making use of functions and routines from these modules are used to conduct our experiments. Currently, there are 2 such scripts: overlap_plots.py and resnik_plots.py. These scripts generate and save the necessary plots (as well as any computed (non-randomized) matrices), along with TeX markup, which you can see in report.tex.
Run these scripts from the command line as such:
python overlap_plots.py "PPIs and GO" "<organisms>" "tex.txt"python resnik_plots.py "PPIs and GO" "<organisms>" "go.obo" "sprot.stripped" "tex.txt"
, replacing <organisms> with the organisms you wish to compute plots for (their corresponding files have to be present in the PPIs and GO directory), separated by spaces. tex.txt can be any text file; the TeX markup will be written to this file. Plots are saved to the plots directory, and numpy matrices are saved to the NumPy files directory.
Note: these scripts currently do not give you the option of not saving the computed matrices, which can be quite large. If this is a problem, feel free to raise an issue; alternatively, you can not include paths to numpy matrices when calling routines from the plotting module.