CGI: a new approach for prioritizing genes by Combining Gene expression and protein-protein Interaction data

---

Details:

This program is intended to implement a new approach to prioritize genes by integrating expression data and protein interaction data. The details of the methodology as well as the correct citation for this program are as follows: Ma XT, Lee HJ, Wang L, and Sun FZ. Bioinformatics (2007, in press)

Code and Executables:

Pre-compiled executables: 

• Windows OS: CGI_prog.rar

After downloading the package, unzip it and copy all files into the folder C:/CGI (you can use other folders, but please make sure the paths in the config file is then updated).

README file (pdf format) contains information about the test, how to use the program and describes input/output files.

Sample Data Sets:

For yeast data

Sample input file: yeast_data.rar (46Mb) which contains the following files

• CGI_config.txt - contains the config file for CGI.exe
• Cycle.txt - contains Spellman's cell cycle data of 6178 genes across 77 conditions.
• MIPS_Kernel_L070.txt - contains the diffusion kernel of yeast protein protein interaction network (4136*4136 matrix).
• cycle_A_I_G.txt - contains the list of 3758 yeast genes appeared in Spellman's data, MIPS data and GO annotation data.
• GO_data.txt - contains the GO annotation of yeast genes.
• informativeGO.txt - contains the informative GO nodes.

Sample output files:

• cellcycleoutput.txt.R - contains the output of our prioritizing method CGI, which is an R script (you have to load it into R environment to get the final output)
• cellcycleoutput.txt.R_rltlist.txt - contains the prioritizing result when taking each individual genes as phenotype
• cellcycleoutput.txt.R_rlt.txt - contains the performance index of our CGI approach and that using expression data only

For human Alzheimer's disease data

Sample input file: real_data.rar (52Mb) which contains the following files

• realCGI_config.txt - contains the config file for realCGI.exe
• Alzheimerdata.txt - contains the phenotype value MMSE and the expression profile of 4703 genes across 31 samples.
• HumanPPI_DF070.txt - contains the diffusion kernel of human protein protein interaction network (4703*4703 matrix).
• allgenes.txt.txt - contains the list of 4703 gene names and phenotype MMSE.
• pheno_name.txt - contains the phenotype MMSE (by this file the program knows that we are proritizing genes against this phenotype MMSE).

Sample output files:

• MMSE_tau070.txt - rank of genes by our new approach CGI as well as using gene expression data only and protein degree only (smaller rank means better)
• MMSE_tau070.xls - mannually curated file of the above result, including known Alzheimer's disease gene information

Please contact us if there are any bugs within our program. 
Copyright © 2005 The University of Southern California. All RIGHTS RESERVED.

---

Created Date: Jan 2007
Last Updated Date: Jan 03, 2007
Contact

 Fengzhu Sun, Ph.D.
 Molecular and Computational Biology Program
 Department of Biological Sciences
 University of Southern California
 Los Angeles, CA, 90089
 (213) 740-2413 (phone)
 (213) 740-2424 (fax)
 Email: fsun AT usc DOT edu