Simple python package to render the holdouts and training datasets of active regulatory regions for models with the task to predict them.

As usual, just download it using pip:

pip install arrp

Since some software handling coverages sometime get slightly different results, here's three of them:

If you don't have it already, you will need to install the package bedtools.

Most commonly you just need to run the following:

sudo apt install bedtools
pip install arrp

From within the repo run:

from arrp import build
build(target="dataset")

Where "dataset" is the path to your dataset. The default one is the one in the repository.

By default it uses hg19, as it is the genome used in the labeled data currently available from the Wasserman team. This is one of the numerous settings available.

Once you have rendered the dataset you can run the following snippets:

from arrp_NNs import cnn
cnn(target="dataset")

from arrp_NNs import mlp
mlp(target="dataset")

from arrp_NNs import mmnn
mmnn(target="dataset")