This repo contains the code to reproduce the experiments of the paper Can deep learning improve genomic prediction of complex human traits?.

• Download and install Tensorflow and Keras.
  • The file requirements.txt contains a simple list of all the packages in the current environment, and their respective versions.
• Download the biobank dataset and traits.
  • Save genotypes
  • Save phenotypes in a csv, each trait should be in a column
• Run a GWAS for all traits and save the results in a csv file
  • This csv should have the p-values of each SNP for each trait (in columns)

• data/ folder contains the GWAS for all the traits used in the paper

python main.py --mlp --trait height

python main.py --cnn --trait height --unif

python main.py --mlp --trait BHMD --unif -k 50000

• see GA directory

• see notebook and R code

MIT