Table 1. Description of the datasets/cases used in this study.

Experimental Application	Dataset	Tumor Type	Array Platform	Hybridization Method	Number of Arrays Used	Phenotype	Reference
I	I	Breast cancer	Spotted cDNA arrays (svcc-8k)	Reference design	39	Estrogen receptor (ER) status	[29]
	II	Breast cancer	Agilent 22K custom oligonucleotide arrays	Reference design	52	Estrogen receptor (ER) status	[28]
	IIIa	HNSCCc	Agilent 4×44K oligonucleotide arrays	Reference design	16	Tumor/Normal
	IVb	HNSCCc	Agilent 4×44K oligonucleotide arrays	Self hybridization	16	Tumor/Normal
II	V (Affymetrix)	UHRR and HBRRd	Affymetrix HG-U133 Plus 2.0 Gene Chip	Reference design	40	Samples A-De	[23]
	VI (Agilent one-color)	UHRR and HBRRd	Agilent Whole Human Genome Oligo Microarray, G4112A	One-color design	20	Samples A-De	[23]
	VII (Agilent two-color)	UHRR and HBRRd	Agilent Whole Human Genome Oligo Microarray, G4112A	Two-color design	10	Samples A,Be	[23]
III	VIII (Microarray)	Breast cancer	UNC custom Agilent two-color microarray	Reference design	8	Basal/Luminal
	IX (RNA-Seq)	Breast cancer	Illumina Genome Analyzer II	RPKM normalized sequence tags	8	Basal/Luminal

^a,bDatasets III and IV contain the same clinical samples that were hybridized using different array designs.

^cHead and Neck Squamous Cell Carcinoma.

^dUniversal Human Reference RNA (UHRR) and Human Brain Reference RNA (HBRR).

^eSample A = 100% UHRR; Sample B = 100% HBRR; Sample C = 75% UHRR, 25% HBRR; Sample D = 25% UHRR, 75% HBRR.