PEAK

PEAK: Integrating Curated and Noisy Prior Knowledge in Gene Regulatory Network Inference.

Web Application to upload your data, predict with NoiseInf, visualize the results:

http://detangle.cs.vt.edu/

PEAK is a framework for predicting gene regulatory network from gene expression data with different types of prior knowledge.

Types of prior knowledge:

1. Noisy: use penaltyScaling option
2. Reliable: use FeatureScaling option

Web Server:

A Web server to submit gene expression data and prior knowledge and to visualize the results will be available soon. http://detangle.cs.vt.edu/

Installation

1- Python requirements:

PEAK works in both Python 2.7+ and Python 3.4+

1- Install the following Python packages:

• scipy
• numpy
• matplotlib
• pandas

You can either use pip or anaconda:

pip install scipy numpy matplotlib pandas

2- build and intsall a modified version of scikit-learn
https://github.com/doaa-altarawy/scikit-learn

Instructions on building scikit-learn can be found here
http://scikit-learn.org/stable/developers/advanced_installation.html

2- R requirements:

Install packages:

• inline
• multicore
• elasticnet
• Matrix
• corpcor
• nnls
• parallel

Detailed instruction how to install R and its required packages:

1- On ubuntu, first install R:

sudo apt-get update
sudo apt-get install r-base
sudo apt-get install r-base-dev

2- Next, install open the R shell:

sudo R

Then in the R shell install packages:

install.packages(c("inline"))
install.packages(c("multicore"))
install.packages(c("elasticnet"))
install.packages(c("Matrix"))
install.packages(c("corpcor"))
install.packages(c("nnls"))
install.packages(c("parallel"))

3- To install the multicore package download it from:
http://cran.r-project.org/src/contrib/Archive/multicore/

Then:

install.packages('/path/to/downloaded/multicore', repos = NULL, type="source")