methCancer-gen: a DNA methylome dataset generator for user-specified cancer type based on conditional variational autoencoder

Table 4 Comparison of cancer type prediction accuracy for SVM classifiers trained based on different dataset

Cancer	Number of testing samples	TCGA only	TCGA & benchmark			TCGA & methCancer-gen
			Number of generated dataset for each cancer type
			100	200	300	100	200	300
BRCA	313	0.796	0.796	0.796	0.796	0.799	0.802	0.809
COAD	102	0.922	0.922	0.922	0.922	0.931	0.951	0.951
GBM	71	0.972	0.972	0.972	0.972	0.972	0.972	0.972
KIRC	45	0.733	0.733	0.733	0.733	0.733	0.733	0.733
LUAD	162	0.969	1.000	1.000	1.000	1.000	1.000	1.000
PAAD	166	0.139	0.168	0.168	0.175	0.398	0.434	0.434
PRAD	20	0.700	0.700	0.700	0.700	0.700	0.700	1.000
SKCM	159	0.868	0.887	0.887	0.887	0.887	0.887	0.931
Average		0.751	0.761	0.761	0.762	0.799	0.809	0.823

ISSN: 1471-2105