This repository is in support of the eADAGE paper. It contains all the code
and necessary data/metadata to repeat all analyses in the paper.
## What is eADAGE?
eADAGE (ensemble ADAGE) is an enhanced version of
[ADAGE](https://github.com/greenelab/adage). It consolidates 100 individual
ADAGE models into one ensemble model. eADAGE automatically extracts
biologically meaningful features from
large-scale transcriptomic data.
## How do I repeat analyses in the eADAGE paper?
The repository is divided into 6 sections in the order of **data_collection**,
**netsize_evaluation**, **ensemble_construction**, **PCA_ICA**,
**node_interpretation**, **medium_analysis**. Each section
provides a shell script that guides the analysis workflow.
* tested on version 2.7.6 and 2.7.9
* Python packages will be installed automatically in the shell script
* required packages: theano, docopt, requests, statsmodels, numpy
* tested on version 3.2.1 and 3.2.3
* R libraries are handled by the library
[pacman](https://github.com/trinker/pacman), please install it first.
* the following libraries will be installed/loaded when necessary by pacman:
affy, affyio, [TDM](https://github.com/greenelab/TDM), doParallel, readr,
ggplot2, gplots, sprint, ff, cluster, plyr, dendextend, gdata, limma
## How should I build a new eADAGE model?
By modifying the *platform* and *organism* parameters in the **data_collection/
data_collection.sh** file, you can build an expression compendium for a new
organism (if the organism uses that one major array platform.).
Then you can follow the instructions in **eADAGE_construction.sh** to build an
eADAGE model for it.
### Contact info
Jie Tan (email@example.com)