Genomic Selection / Phenotypic Prediction Using Neural Networks

This python code leverages the pybrain and scikit-learn libraries, as well as a built-in artifical neural network implementation to perform genomic selection on genotypic and phenotypic data. It compares prediction accuracy between modeling methods.

The purpose of this codebase is to evaluate the predictive performance of neural networks and other alternate statistical modeling techniques when compared to

1. Standard mixed linear models
2. Published alternative models

Running the code

This is intended to be run using a python virtualenvironment on Linux. Set up a virtualenvironment by running the script below.

sudo pip install virtualenvwrapper
source $(which virtualenvwrapper.sh)
mkvirtualenv genomic-neuralnet
pip install -r requirements.txt

To stop working on the code and resume using your system's python executable, deactivate the virtualenvironment using the deactivate command.

deactivate

To continue working on the code, simply say that you wish to work in the genomic-neuralnet virtualenvironment again.

workon genomic-neuralnet