Amogel: a multi-omics classification framework using associative graph neural networks with prior knowledge for biomarker identification

Table 1 Summary of the multi-omics datasets

Dataset	Subtypes and number of samples after data preprocessing	Number of features in mRNA, DNA Methylation and miRNA	Number of features after data preprocessing^a
BRCA	Normal-like: 115 Basal-like: 131 HER2-enriched: 46 Luminal A: 546 Luminal B: 147	20531, 20106, 503	1000, 1000, 502
KIPAN	KICH: 66 KIRC: 318 KIRP: 273	20531, 20111, 472	2000, 2000, 471

^aTo ensure the fair comparison, ANOVA-F filtering threshold for BRCA is 1000 and KIPAN is 2000, same threshold reported in MOGONET [5] and MOGLAM [11]

ISSN: 1471-2105