Table 1.

The overall accuracies (%) and AUCs of our DCNN models for multi-class classification. CNA_DCNN and Gene_DCNN are based on the architecture of Fig. 2 for CNA data and gene expression data, respectively. DCNN_Concat, DCNN_Siamese and DNN_SE are DNN models based on the network architectures described in Fig. 3, Fig. 4, Fig. 5, respectively.

Model (all genes)	Datasets	Performance Measurement
		Accuracy (%)	AUC
CNA_DCNN	CNA	50.5	0.677
Gene_DCNN	Gene expression	77.3	0.832
DCNN_Concat	CNA and gene expression	79.2	0.850
DCNN_Siamese	CNA and gene expression	76.7	0.838
DNN_SE	CNA and gene expression	77.3	0.838