Table 1.

Evaluation data sets

Data Set	Array Type	Probes	Regions	Cancer Type	Groups (Samples)
Chin et al.	spotted oligo	26,755	223	breast	ER+ (113) vs. ER- (57)
Douglas et al.	BAC	3,032	142	colorectal	MSI (7) vs. CIN (30)
Fridlyand et al.	BAC	1,877	231	breast	TP53+ (10) vs. TP53- (52)
Myllykangas et al.	cDNA	11,342	260	gastric	diffuse (15) vs. intestinal (23)
Nymark et al.	cDNA	10,953	242	lung	asbestos-exposed (11) vs. non-exposed (9)
Postma et al.	spotted oligo	26,755	111	colorectal	good (16) vs. bad response (16)
Smeets et al.	BAC	4,196	143	head and neck	HPV+ (12) vs. HPV- (12)
Wrage et al.	spotted oligo	25,549	23	lung	BM+ (13) vs. BM- (15)
Simulation 0	in-situ oligo	42,331	440		(15) vs. (15)
Simulation 5	in-situ oligo	42,331	489		(15) vs. (15)
Simulation 10	in-situ oligo	42,331	525		(15) vs. (15)

Eight public data sets were collected to evaluate the performance of CGHpower. They represented five different cancer types and BAC, cDNA and oligo-based microarray platforms, with resolutions varying from 2 K to 27 K array elements. The last column contains the distinguishing factor used to divide the data set into two groups, along with the number of arrays in each group. The simulated data sets were generated by introducing artificial aberrations into a set of clinical genetics samples. A total of 11 simulations were generated, and the remaining ones are available at http://www.cangem.org/cghpower/. ER = estrogen receptor, MSI = microsatellite instability, CIN = chromosomal instability, HPV = human papilloma virus, BM = bone marrow metastasis.