Skip to main content

Table 2 Mean results across tenfold cross-validation of prediction for each investigated species with the 3 different input data type: adjacency matrices only, adjacency matrices concatenated with embedding vector-based matrices, embedding vector-based matrices only

From: Efficient link prediction in the protein–protein interaction network using topological information in a generative adversarial network machine learning model

Species

Adjacency only

Combined

Embedding only

AUROC

AUPRC

NDCG

AUROC

AUPRC

NDCG

AUROC

AUPRC

NDCG

Homo sapiens

0.913

0.169

0.761

0.915

0.179

0.767

0.761

0.022

0.606

Saccharomyces cerevisiae

0.931

0.202

0.781

0.930

0.210

0.787

0.789

0.034

0.635

Mus musculus

0.909

0.137

0.742

0.904

0.135

0.739

0.739

0.020

0.599

Rattus norvegicus

0.925

0.252

0.809

0.923

0.258

0.812

0.750

0.027

0.633

Sus scrofa

0.898

0.120

0.721

0.895

0.126

0.729

0.745

0.017

0.583

Mean

0.915

0.176

0.763

0.913

0.182

0.767

0.757

0.024

0.612

  1. AUROC: area under the receiver operating characteristic curve; AUPRC: area under the precision-recall curve; NDCG: normalized discounted cumulative gain; Combined: adjacency matrices concatenated with embedding vector-based matrices as input