A pan-cancer somatic mutation embedding using autoencoders

BMC Bioinformatics

Table 1 Classification results for 40 cancer subtypes

Primary site	Project name	Test samples	AUC latent	AUC De.	AUC Nd.
Head and neck	ORCA-IN	36	0.75	0.50	0.55
Brain	LGG-US	56	0.81	0.62	0.80
Blood	CLLE-ES	102	0.83	0.84	0.76
Head and neck	THCA-SA	28	0.82	0.66	0.81
Liver	LINC-JP	79	0.54	0.50	0.50
Lung	LUSC-KR	34	0.77	0.57	0.50
Skin	GACA-CN	24	0.79	0.50	0.50
Stomach	LICA-FR	49	0.67	0.50	0.51
Blood	ESCA-CN	65	0.77	0.5	0.61
Lung	EOPC-DE	40	0.81	0.74	0.55
Colorectal	COCA-CN	63	0.60	0.50	0.51
Skin	SKCM-US	67	0.84	0.50	0.50
Liver	LICA-CN	79	0.61	0.50	0.50
Blood	ALL-US	15	0.83	0.91	0.94
Skin	SKCA-BR	20	0.84	0.50	0.50
Brain	GBM-US	55	0.87	0.50	0.66
Nervous system	NBL-US	18	0.96	0.95	0.96
Blood	LAML-KR	41	0.97	0.73	0.64
Prostate	PRAD-UK	28	0.65	0.5	0.50
Prostate	PRAD-US	51	0.83	0.58	0.74
Blood	MALY-DE	49	0.82	0.50	0.50
Kidney	KIRC-US	82	0.90	0.50	0.58
Brain	PBCA-DE	90	0.79	0.77	0.59
Kidney	RECA-EU	49	0.84	0.5	0.5
Blood	AML-US	18	0.94	0.96	0.96
Breast	BRCA-UK	28	0.59	0.50	0.50
Kidney	KIRP-US	33	0.90	0.50	0.57
Prostate	PRAD-CA	58	0.88	0.71	0.50
Stomach	STAD-US	58	0.84	0.50	0.50
Stomach	GACA-JP	115	0.80	0.50	0.50
Liver	LIRI-JP	32	0.88	0.50	0.75
Breast	BRCA-US	189	0.75	0.60	0.50
Lung	LUSC-US	39	0.91	0.50	0.50
Esophagous	ESAD-UK	61	0.90	0.50	0.50
Colorectal	COAD-US	51	0.83	0.50	0.50
Breast	BRCA-FR	15	0.82	0.50	0.56
Pancreas	PACA-AU	73	0.72	0.60	0.50
Pancreas	PACA-CA	54	0.87	0.50	0.89
Head & Neck	THCA-US	76	0.85	0.85	0.50
Breast	BRCA-EU	114	0.79	0.56	0.50

The number of the test samples for the corresponding class is detailed. Area under the Roc curve is detailed for classifiers on Latent Space, Deleterious and Non-Deleterious input data. Tumor subtypes where the classification performance is improved in the latent space are highlighted in bold

ISSN: 1471-2105