Methods for training and interpretation of an ensemble of neural networks for multi-task functional prediction of accessibility or histone modifications from DNA sequence.
These are instructions for how to run the packaged version of DeepAccess training and interpretation. We provide a tutorial for running DeepAccess training and interpretation as a google colab notebook. You may also download deepaccess which has been tested for Ubuntu 18.04.3 (will not work on Mac or Windows) as a complete binary using zip or tarball on our github releases page.
To run DeepAccess with regions (bedfile format) you must install bedtools and add it to your path. Bedtools binaries are available here. After installation, you can add bedtools to your path via the terminal or modifying your ~/.bashrc
export PATH="/path/to/bedtools:$PATH"
If you choose to install deepaccess through PyPI, the default will install a version of tensorflow for GPU usage. If you do not intend to use GPUs, please first install tensorflow for cpu:
pip install tensorflow-cpu
DeepAccess can be installed as a command line tool with pip:
pip install deepaccess
or with bioconda:
conda install -c bioconda deepaccess
If deepaccess is properly installed, you should be able to run the following without errors:
deepaccess -h