The purpose of this utility is to take raw, heterogenous gene expression data (GSE files) from GEO and output a normalized matrix, with genes as columns and experiments as rows, for conducting gene expression meta-analyses such as described in some of our papers:
- R language (tested on 2.13.0), including RScript, on your PATH
- Python 2.x, also on PATH
- GNU Make and GNU grep (needs to support -P option)
- ~150 GB free disk space
First, download GSEs from NCBI's FTP servers, GEO and Entrez Gene metadata, and R package dependencies by running:
make download
Alternatively if you've already downloaded the GSEs, you can symlink
data/GSE
to a flat directory containing the GSEs and download the
other dependencies individually (see the Makefile).
Now make the matrix for the species of your choice by running, e.g.:
make species-matrix SPECIES="Caenorhabditis elegans"
or for the platform of your choice:
make platform-matrix PLATFORM="GPL200"
being sure to surround the species/platform of interest by quotes. The resulting matrices (raw and quantile normalized) will be output into the data/ directory.
Copyright (C) 2011 Oklahoma Medical Research Foundation
Distributed under the Eclipse Public License.