DeepERA: deep learning enables comprehensive identification of drug-target interactions via embedding of heterogeneous data

Prerequists (tested version):

• Python (3.6.8)
• pandas (0.21.1)
• pytorch (1.2.0)
• rdkit (2019.09.3.0)
• scikit-learn (0.20.3)
• numpy (1.16.1)

We recommend installing the above packages with the tested version to avoid any potential errors/warnings due to version changes.

Installation with conda environment:

1. Create environment with Python 3.6.8:

conda create -n DeepERA python=3.6.8

2. Install packages in the DeepERA environment

conda activate DeepERA
conda install -c conda-forge scikit-learn=0.20.3
conda install -c rdkit rdkit=2019.09.3.0
conda install pandas=0.24.1
conda install pytorch==1.2.0 torchvision==0.4.0 cudatoolkit=10.0 -c pytorch
conda install numpy=1.16.1

The installation of pytorch depends on the cuda version. If your cuda version is 9.2, please use the following command instead:

conda install pytorch==1.2.0 torchvision==0.4.0 cudatoolkit=9.2 -c pytorch

Please refer to the instruction page of pytorch installation if non-linux platform (e.g. MacOS, Windows) is used.

Run the code

1. Process raw data and processed data are stored in the folder Processed_data. Please replace the files in the folder Raw_data to run your own task:

python preprocess_data_DeepERA.py

2. Run training and prediction for the data in the folder Processed_data:

bash run_DeepERA.sh

Citation

Please cite the following paper:

• Le Li, Shayne D. Wierbowski, Haiyuan Yu. DeepERA: deep learning enables comprehensive identification of drug-target interactions via embedding of heterogeneous data. Manuscript in submission.

Contact

Please contact me ([email protected]) for any questions.