Peter Sadowski and Michael Zeller (Team IGB) submission and scripts for sbv IMPROVER 2013 subchallenge 2
Clone the git repository and also run the following to get the
git submodules pylearn2 and cybtpy.
git submodule init
git submodule update
You will also need to extract the competition provided data from the data directory, for example:
cd data
unzip *.zip
After extracting, copy the files from the SBV_STC_subchallenge1_gold_standard folder
into the SBV_STC_subchallenge2 folder.
Code has been tested on CentOS 6 using Python 2.7.6 and R 3.0.1.
Required R packages are gplots and VGAM.
The neural network training is performed using the Pylearn2 submodule, with the script and readme located in
improver2013/opt/pylearn2/pylearn2/scripts/improver2013/
The pre- and post-analysis scripts for the competition are in the scripts dir, and
they work with files from the data and out directories.
process-microarray.r: Processes the GEx and P data and includes the
batch number to construct the training data as a single file.
transform-data.r: Implements the transformations of the GEx data to
be clipped from 0 to 4.
convert-to-significance-subchallenge2.r: Creates the submission file
and figures using combinations of t-tests (bayesian, standardard),
transformations (fisher z, logistic), and combinations of training
data (using GEx, rat P only, rat and human P without GEx).