Drug repurposing for schizophrenia through Deep Graph Networks

This repository provides a reference implementation of our paper Controlling astrocyte-mediated synaptic pruning signals for schizophrenia drug repurposing with Deep Graph Networks.

Requirements

Note: we assume Miniconda/Anaconda is installed, otherwise see this link for correct installation. The proper Python version is installed during the first step of the following procedure.

1. Install the required packages and create the environment

   • conda env create -f env.yml

2. Activate the environment

   • conda activate schizophrenia

Experiments

• To rebuild the dataset run:

  python3 preprocessing.py

• To perform the model selection with k-fold cross validation run:

  python3 main_model_selection.py

• To perform risk assessment of the model on the held-out test set run:

  python3 main_risk_assessment.py

• To perform the prediction of a new set of molecules, i.e., for biological validation purposes, run:

  1. python3 preprocessing.py --bioval --source_path <source_path_of_data>
  2. python3 main_bioval_prediction.py --model_path <saved_model_path> --bioval_data_path <data_path>