Table 3 Results of the proposed and baseline models with 700 genes for molecular subtype classification on the TCGA pan-cancer dataset and cancer subtype classificaiton on the TCGA BRCA dataset
From: A multimodal graph neural network framework for cancer molecular subtype classification
Model | Pan-cancer | BRCA | ||
|---|---|---|---|---|
Accu.a | F1 | Accu.a | F1 | |
Proposed w/ GAT | 83.9% ± 1.4% | 0.84 ± 0.01 | 86.4% ± 1.9% | 0.87 ± 0.02 |
Proposed w/ GCN | 81.2% ± 0.6% | 0.81 ± 0.01 | 83.8% ± 0.9% | 0.84 ± 0.01 |
FC-NN | 78.4% ± 0.8% | 0.75 ± 0.02 | 80.8% ± 1.1% | 0.80 ± 0.02 |
GCN (Original) [23] | 77.6% ± 0.9% | 0.76 ± 0.02 | 82.8% ± 1.2% | 0.84 ± 0.01 |
GCN (Modified) | 78.5% ± 1.2% | 0.77 ± 0.02 | 81.8% ± 1.4% | 0.82 ± 0.01 |
Multi-omics GCN (Original) [1] | 78.6% ± 0.9% | 0.78 ± 0.01 | 81.8% ± 1.1% | 0.82 ± 0.01 |
Multi-omics GCN (Modified) | 80.2% ± 0.8% | 0.79 ± 0.01 | 82.8% ± 0.9% | 0.83 ± 0.01 |
GrAMME (Modified) [25] | 81.4% ± 1.3% | 0.81 ± 0.03 | 82.8% ± 1.6% | 0.84 ± 0.03 |
Multi-omics GAT (Original) [7] | 76.3% ± 1.2% | 0.76 ± 0.02 | 81.8% ± 1.3% | 0.82 ± 0.02 |
Multi-omics GAT (Modified) | 79.7% ± 1.3% | 0.79 ± 0.02 | 82.8% ± 1.4% | 0.84 ± 0.02 |