Data for ProKG model: Extensible Knowledge Graph-Guided Prediction of Compound-Protein Interactions

File Description ProKG_env.tar.gz Packaged Conda environment containing the Python runtime and dependencies required to run ProKG. raw.tar.gz Raw and cleaned BioLink knowledge graph data, including edges, node mappings, node features, and source datasets. feature.tar.gz Feature extraction modules and supporting resources for generating compound and protein representations. downstream.tar.gz Processed BioLink data, samples, features, and graph inputs used for downstream model training and evaluation. pre.tar.gz Processed input data and resources used for prediction, external validation, and independent benchmark evaluation. models.tar.gz Trained downstream ProKG model checkpoints, including models generated from different folds and experimental settings. pre_trained_model.tar.gz Pre-trained graph model checkpoints used to initialize downstream ProKG tasks.