Skip to main content

Table 2 Performance of cancer type-specific driver gene prediction

From: Identifying cancer driver genes based on multi-view heterogeneous graph convolutional network and self-attention mechanism

Types of cancer

AUC

AUPRC

LUAD

MOGONET

0.8960 ± 0.0016

0.6106 ± 0.0074

Multi-omics fusion

0.8904 ± 0.0007

0.6221 ± 0.0039

RGCN

0.8720 ± 0.0017

0.4419 ± 0.0095

GCN

0.8042 ± 0.0014

0.4187 ± 0.0065

GAT

0.8180 ± 0.0018

0.3398 ± 0.0055

EMOGI

0.8709 ± 0.0019

0.5591 ± 0.0105

MTGCN

0.9019 ± 0.0012

0.6279 ± 0.0099

MRNGCN

0.9427 ± 0.0011

0.6287 ± 0.0117

BRCA

MOGONET

0.8944 ± 0.0008

0.6350 ± 0.0032

Multi-omics fusion

0.9050 ± 0.0006

0.6714 ± 0.0030

RGCN

0.8557 ± 0.0029

0.3983 ± 0.0148

GCN

0.8813 ± 0.0009

0.5866 ± 0.0058

GAT

0.8673 ± 0.0044

0.4387 ± 0.0136

EMOGI

0.8989 ± 0.0007

0.6482 ± 0.0045

MTGCN

0.9061 ± 0.0006

0.6583 ± 0.0040

MRNGCN

0.9120 ± 0.0008

0.6920 ± 0.0041

BLCA

MOGONET

0.9368 ± 0.0005

0.5884 ± 0.0111

Multi-omics fusion

0.9482 ± 0.0008

0.6539 ± 0.0153

RGCN

0.8420 ± 0.0023

0.4404 ± 0.0095

GCN

0.8712 ± 0.0013

0.3191 ± 0.0091

GAT

0.8929 ± 0.0007

0.2892 ± 0.0068

EMOGI

0.9359 ± 0.0005

0.5485 ± 0.0101

MTGCN

0.9495 ± 0.0005

0.6568 ± 0.0077

MRNGCN

0.9544 ± 0.0005

0.7248 ± 0.0052

LIHC

MOGONET

0.8739 ± 0.0023

0.4306 ± 0.0198

Multi-omics fusion

0.8900 ± 0.0017

0.5220 ± 0.0082

RGCN

0.8648 ± 0.0024

0.5016 ± 0.0133

GCN

0.8098 ± 0.0019

0.3017 ± 0.0090

GAT

0.8472 ± 0.0014

0.2340 ± 0.0042

EMOGI

0.8753 ± 0.0023

0.3845 ± 0.0193

MTGCN

0.8937 ± 0.0016

0.4645 ± 0.0195

MRNGCN

0.9109 ± 0.0012

0.5468 ± 0.0149

  1. Bold values indicates the best performance
  2. AUC value and AUPRC comparison of our method MRNGCN and other comparison methods on specific cancer data sets, namely LUAD, BRCA, BLCA and LIHC