Codes for reproducing the submission for Team Lavawizard in 2019 ENCODE Imputation Challenge

Installation

1. git clone https://github.com/ccchang0111/ENCODE_imputation_2019.git
2. conda env create -f env.yml

How to train the model & generate predictions

1. Download all original training & validation data (.bigwig) into /data/training_data/ and /data/validation_data/ respectively
2. Download submission_template.bigwig file to the folder /data/submission_template/
3. Run Lavawizard_pipeline.sh from the command line

or alternatively,

How to generate predictions from pretrained models

1. Download all pretrained models (23 files) to the folder /data/processed_data/keras_model/
2. Run Lavawizard_pipeline_using_pretrained_model.sh