Integrating heterogeneous genomic data to accurately identify disease subtypes

BMC Medical Genomics

Table 2 Performance comparison of different integrative methods on lung and kidney cancer datasets produced by TCGA

Lung	Methylation	mRNA	miRNA	Concatenation	SNF	iBFE
PCC_intraclass	0.20	0.10	0.11	0.15	0.12	0.65
PCC_interclass	−0.13	−0.05	−0.06	−0.10	−0.06	−0.39
PCC_intraclass-PCC_interclass	0.33	0.15	0.17	0.25	0.17	1.04
Dist_intraclass	160.41	142.37	22.57	218.65	0.02	2.81
Dist_interclass	239.13	157.70	26.40	287.85	0.02	6.29
Dist_interclass/Dist_intraclass	1.49	1.11	1.17	1.32	1.18	2.24
ACC_rfLOO	0.99	0.97	0.97	1.00	0.98	1.00
Kidney
PCC_intraclass	0.10	0.15	0.11	0.07	0.16	0.37
PCC_interclass	−0.05	−0.07	−0.05	−0.04	−0.12	−0.19
PCC_intraclass-PCC_interclass	0.15	0.22	0.16	0.11	0.27	0.56
Dist_intraclass	207.37	164.80	22.90	277.16	0.02	3.48
Dist_interclass	226.39	193.05	25.64	295.57	0.02	4.97
Dist_interclass/Dist_intraclass	1.09	1.17	1.12	1.07	1.52	1.43
ACC_rfLOO	0.98	0.98	0.96	0.93	1.00	0.95

The best performer was highlighted with the darkest color
PCC_intraclass: average Pearson correlation coefficients of patients within the same classes; PCC_interclass: average Pearson correlation coefficients of patients from different classes; Dist_intraclass: average Euclidean distance of patients within the same classes;Dist_interclass:average Euclidean distance of patients from different classes;ACC_rfLOO: accuracy of leave-one-out cross-validation by random forest based on the clustering labels by k-means

ISSN: 1755-8794