Table 5 Median WAF scores for the ontology ablation studies. Comparison of the best knowledge graph embeddings methods RDF2Vec and OPA2Vec with Random Forest or XGB for the knowledge graph HPs+GO+LD. Results that are statistically significantly different when compared to HPs+GO+LD are underlined. Best results in bold
From: Multi-domain knowledge graph embeddings for gene-disease association prediction
Random Forest | XGB | ||||
|---|---|---|---|---|---|
RDF2vec | OPA2vec | RDF2vec | OPA2vec | ||
HPs+GO+LD | 0.749 | 0.770 | 0.737 | 0.768 | |
GO Ablation | HPs+GO(-BP)+LD | 0.747 | 0.763 | 0.740 | 0.759 |
HPs+GO(-CC)+LD | 0.744 | 0.767 | 0.765 | 0.765 | |
HPs+GO(-MF)+LD | 0.749 | 0.770 | 0.748 | 0.763 | |
HPs+GO(Only BP)+LD | 0.742 | 0.769 | 0.730 | 0.767 | |
HP Ablation | HPs+GO(-F)+LD | 0.720 | 0.740 | 0.713 | 0.734 |
HPs+GO(-PA)+LD | 0.553 | 0.564 | 0.562 | 0.567 | |
HPs+GO(-MI)+LD | 0.723 | 0.738 | 0.716 | 0.738 | |
HPs+GO(-CC)+LD | 0.712 | 0.733 | 0.713 | 0.731 | |
HPs+GO(-CM)+LD | 0.731 | 0.736 | 0.714 | 0.733 | |
HPs+GO(Only PA)+LD | 0.743 | 0.735 | 0.742 | 0.734 | |