Knowledge Graph Embeddings and Chemical Effect Prediction, 2020.

Scripts and outputs related to the paper Prediction of Adverse Biological Effects of Chemicals Using Knowledge Graph Embeddings.

A snapshot of TERA (the Toxicological Effect and Risk Assessment Knowledge Graph) is avalible as a Zenodo dataset:

To run the codes you must install KGE-Keras.

To reproduce results from paper run:

bash run.sh

This will take several days, depending on hardware avalible. 3-5 days on Nvidia GTX 1070ti. Set --MAX_TRIALS 0 to use default parameters.

Alternativly, unzip results.zip and run

python3 analyse_results.py

Related publications

Erik B. Myklebust, Ernesto Jimenez Ruiz, Jiaoyan Chen, Raoul Wolf, Knut Erik Tollefsen. Prediction of Adverse Biological Effects of Chemicals Using Knowledge Graph Embeddings. Accepted for publication in the Semantic Web Journal, 2021. (arXiv) (Paper) (REPOSITORY)

Name		Name	Last commit message	Last commit date
Latest commit History 10 Commits
models		models
om_outputs		om_outputs
reference_mappings		reference_mappings
LICENSE		LICENSE
README.md		README.md
analyse_embeddings.py		analyse_embeddings.py
analyse_kg.py		analyse_kg.py
analyse_mappings.py		analyse_mappings.py
analyse_results.py		analyse_results.py
create_data.py		create_data.py
data.zip		data.zip
results.zip		results.zip
run.sh		run.sh
run_all_models.py		run_all_models.py
split_taxonomy.py		split_taxonomy.py
string_matcher.py		string_matcher.py