Table 1 Performance of DL-PPI on three datasets in relation to comparative methods
From: DL-PPI: a method on prediction of sequenced protein–protein interaction based on deep learning
Dataset | Partition scheme | Method | ||||
|---|---|---|---|---|---|---|
DNN-PPI | TAGPPI | PIPR | GNN-PPI | DL-PPI | ||
SHS27k | Random | 72.06 | 85.46 | 84.28 | 87.35 | 89.12 |
BFS | 50.26 | 49.68 | 47.39 | 68.67 | 72.95 | |
DFS | 59.43 | 63.57 | 54.25 | 71.82 | 78.07 | |
SHS148k | Random | 87.26 | 89.21 | 91.04 | 90.07 | 92.49 |
BFS | 56.44 | 55.9 | 59.87 | 67.42 | 68.87 | |
DFS | 59.18 | 67.35 | 62.66 | 84.05 | 85.45 | |
STRING | Random | 82.04 | 89.03 | 92.76 | 93.61 | 94.85 |
BFS | 57.89 | 58.93 | 57.15 | 76.85 | 77.53 | |
DFS | 59.52 | 68.04 | 65.48 | 90.38 | 92.76 | |