Michael Mooney, PhD
ABOUT RESEARCH CODE RESOURCES

PolyGA:  Quick Start Guide

(Back to PolyGA)

This guide will walk you though the basic procedure for running PolyGA with the example data provided.
To start, change to the directory where you downloaded the PolyGA archive. Extract the contents and then change to the program directory:

tar -xzvf polyga-1.1b.tar.gz
cd polyga-1.1b

Now, use the -h option to see all available command-line options for the PolyGA program:

python polyga.py -h

Next, view the supplied configuration file (polyga.conf) so that you are familiar with the various program parameters. By default, the program will perform a short search (50 generations) for pairs of interacting SNPs.

cat polyga.conf

Next, view the R script that will be used to perform the statistical test of association for each candidate feature set. By default, the scan.glm.2D() function from the GenABEL R package will be used to test for SNP interactions. You will need to install the GenABEL R package before proceeding (http://www.genabel.org/packages/GenABEL). The provided R script is also setup to use the SNOW and foreach R packages to parallelize the statistical calculation and speed-up the run-time. If you do not want to parallelize the R calculations, you can modify the R script. For more details see the user manual.

cat polyga_Rscript_GenABEL.r

Now you are ready to run a short search using the supplied configuration file, network data and simulated genotypes. The results from this search will be saved in the current directory in a file named "test1_out.h5". Note that by default the search is done at the feature level, so the -a option is not strictly necessary in this case.

python polyga.py -a feature -c polyga.conf -g ./data/ensembl_genes.txt \
-i ./data/interactions.txt -f ./data/snp_map.txt -o test1

To view the results, use the -r option and the -t option to specify the results file and p-value threshold. Any SNP pairs with p-values less than or equal to 0.05 will be printed to a file named "test1_out.assoc". Note that since the search is stochastic you may have to adjust the threshold value.

python -r test1_out.h5 -t 0.05

evo·comp·bio