/ENCODE-E2G

Python package to predict enhancer-gene interactions supervised on CRISPRi data

Primary LanguageJupyter NotebookMIT LicenseMIT

ENCODE-E2G

Install requirements

conda create --name <env> --file requirements.txt

Run

You need three files to run ENCODE-E2G models:

- TSS file
- CRISPRi E-G (enhancer-gene) dataset
- Feature table

tss contains the TSS file we have used. crispri has the CRISPRi datasets for training the ENCODE-E2G and ENCODE-E2G_Extended models. Feature tables are binary matrices which specify the features to be used in each model. We have included the the full and ablated models of ENCODE-E2G and ENCODE-E2G_Extended in feature_table.

This demo file shows step-by-step how to run the ENCODE-E2G models, save them, predict the CRISPRi E-G pair, plot the analysis figures, and perform the genome-wide predictions for the provided E-G pairs for every ENCODE cell types.