This project focuses on enhancing drug discovery through the use of knowledge graph modeling. By predicting likely missing links in a directed heterogeneous multigraph medication-gene network, we aim to improve the explainability and effectiveness of drug discovery processes.
- Predict the likely missing links in Hetionet KG biology network.
- Improve drug discovery and explainability by leveraging advanced knowledge graph embeddings.
Python
Libraries and Frameworks: PyKeen
Models: TransE, TransH, RotatE
git clone https://github.com/arushi-08/Drug-Discovery-Knowledge-Graphs.git
cd Drug-Discovery-Knowledge-Graphs
Install depedencies
pip install -r requirements.txt
Run the TransE KGE Experiments (note: TransH and RotatE experiments are triggered in similar fashion, however these models are bigger):
python train_transe_hetionet.py