PheKnowLator Human Disease Knowledge Graph Benchmarks Embeddings -- v1.0.0

PKT Human Disease Knowledge Graph Benchmark Builds - Embeddings (v1.0.0)

Build Date: September 03, 2019

Please note that all resources linked below redirect to a publicly Google Cloud Storage bucket where all data are publicly accessible. Routing users from this wiki page is perfectly safe and allows us to avoid requiring users to have a Google account and login to download data. If you have any questions or concerns, please email the project maintainer at callahantiff@gmail.com.

The KG Benchmark Builds can also be downloaded from Zenodo:
👉 KGs: https://doi.org/10.5281/zenodo.7030200
👉 Embeddings: https://zenodo.org/record/7030189

A modified version of the DeepWalk algorithm was implemented to generate molecular mechanism embeddings from the biomedical knowledge graph. A t-SNE plot of the dimensionality reduced mechanism embeddings is shown in Figure. For this release, the hyperparameters were set to 512 dimensions, 100 walks, walk length of 20, and a window of 10. Two types of KGs were embedded: (1) the full KG; and (2) the full KG with deductive closure using the OWL 2 EL reasoner, ELK via Protégé v5.1.1. ELK is able to classify instances and supports inferences over class hierarchies and object properties. inference over disjointness, intersection, and existential quantification (ontology class hierarchies).