Table 4. Detailed description of the dataset used for performance evaluation.
For all datasets, we used the AD-gene network published by the IntAct Molecular Interaction Database, which is curated by broad literature searches. However, since the size of the IntAct(AD) was small, interactome data were integrated to increase the size of the training dataset.
| Dataset ID | Description of dataset | Sample Size (number of interactions) |
||
|---|---|---|---|---|
| Interactome dataset | Use of AD-related genes (seed gene) |
Normal (class label 0) |
AD (class label 1) |
|
| 1 | IntAct(AD) + bPPI | Y | 3,241 | 3,241 |
| 2 | IntAct(AD) + bPPI + HumanNet (5%) | Y | 4,916 | 4,916 |
| 3 | IntAct(AD) + bPPI + HumanNet (10%) | Y | 7,013 | 7,013 |
| 4 | IntAct(AD) + bPPI | N | 23,546 | 23,546 |
| 5 | IntAct(AD) + bPPI + HumanNet (5%) | N | 46,206 | 46,206 |
| 6 | IntAct(AD) + bPPI + HumanNet (10%) | N | 69,296 | 69,296 |