A benchmark study of deep learning-based multi-omics data fusion methods for cancer

Table 4 The sizes and features of cancer benchmark datasets used in classification task

Cancers	Categories (cancer subtypes)	# of samples	# of features
Cancers	Categories (cancer subtypes)	# of samples	Exp, Meth, miRNA
BRCA	Luminal A: 28, Luminal B: 15, Basal-like: 12, HER2-enriched: 4	59	6000, 5000, 892
GBM	Proneural: 71, Classical: 70, Mesenchymal: 84, Neural: 47	272	6000, 5000, 534
SARC	DDLPS: 50, LMS: 80, UPS: 44, MFS: 17, MPNST: 5, SS: 10	206	6000, 5000, 1046
LUAD	TRU: 51, PI: 52, PP: 41	144	6000, 5000, 554
STAD	EBV: 20, MSI: 38, GS: 43, CIN: 97	198	6000, 5000, 519

# number, Exp gene expression, Meth DNA methylation, miRNA miRNA expression, DDLPS dedifferentiated liposarcoma, LMS leiomyosarcoma, UPS undifferentiated pleomorphic sarcoma, MFS myxofibrosarcoma, MPNST malignant peripheral nerve sheath tumor, SS synovial sarcoma, TRU formerly bronchioid, PI formerly squamoid, PP formerly magnoid, EBV Epstein–Barr virus, MSI microsatellite instability, GS genomically stable, CIN chromosomal instability

ISSN: 1474-760X