README.md

paper deep learning vs gblup

Datasets and scripts used for the article "Deep learning vs GBLUP for whole-genome predictions: relative efficiency with additive and non-additive continuous phenotypes"

• deep_learning_model_run.ipynb is the Python notebook with the code to run the deep learning model developed in the paper. This is tensorflow/keras implementation, which can be run on Google Colaboratory (or a similar Jupyter Notebook engine)
• the folder support_scripts contains all Python dependencies needed to run the DL model (download and prepare the data, build the model, parse and collect results)
• the data are stored in this repository (simulated phenotypes) and on zenodo.org (kinship matrices)
• the R script to simulate the phenotypes with different relative contributions of additive, dominant and epistatic genetic effects is also included (in support_scripts)