Data Mining and Information Systems Laboratory
Laboratory of Prof. Jaewoo Kang

  COSSY :   Context-Specific Subnetwork discoverY

COSSY is an algorithm to discover important subnetworks differentiating between two phenotypes (context). It automatically finds differentially expressed subnetworks of closely interacting molecules from molecular interaction networks (such as KEGG or STRING) using gene expression profiles. This is the first non-greedy approach of its kind. COSSY works for any interaction network regardless of the network topology. It can also be used as a highly accurate classification platform. COSSY has been implemented in R.

1. Program availability and requirements

  • Operating systems : Windows, Linux, Mac
  • R version :  2.15.3 or later
  • R Packages: utils, stats, limma, igraph (>= 0.6.5)

2. Installation

  • Install R (skip this step if you have R 2.15.3 or later)
  • Download the following files:
    1. cossy R package (v.1.0) : cossy_1.0.tar.gz (21KB)
    2. cossy data : (18.2MB)
  • Run R.
  • Unzip cossy data ( inside the working directory. You can get/set the working directory by the following commands.
    getwd() # shows the currect working directory
    setwd("my_dir")	# set the working directory to "my_dir"
  • Install limma package.
  • Install igraph package
    install.packages("igraph", repos="")
  • Install cossy package
    install.packages(pkgs="cossy_1.0.tar.gz", repos=NULL, type="source") # Put the path to the cossy package file in 'pkgs' parameter
  • Load libraries
  • Example code to run
    mis <- cossy(dataset="leukemia", network="kegg")
    pred <- cossy(dataset="prostate3", network="string", T=5, loocv=TRUE)	#Note: It may take a long time to finish.

3. Documentation

Use the following commands to see the R documentation.


4. Souce Code

Source Code :

5. License

COSSY is licensed under the GNU General Public License and is 100% freely available to both commercial and academic users. For the full text of the license:>

6. Contact - for bugs, comments and questions

Ashis Saha:
Jaewoo Kang:>

Last updated on July 23, 2013