Microarray Normalization Revisited for Reproducible Breast Cancer Biomarkers

Table 4

Impact of data processing pipelines on 6 algorithms for breast cancer subtype classification: Percent misclassification for pairwise comparisons between pipelines.

		Breast cancer subgroup prediction: percent misclassification
Algorithm	#	A	B	C	D	E	F	G	H	I	J	K	L	M	N
scmgene	4	49.9	31.1	58.5	9.7	59.6	1.1	8.1	10.8	6.2	4.0	5.8	23.6	11.1	12.0
scmod1	4	21.5	18.2	16.1	13.2	20.7	2.0	6.1	18.4	2.5	4.3	2.4	23.4	18.1	17.9
scmod2	4	22.5	20.0	16.8	11.3	18.7	1.7	5.1	18.1	1.6	5.3	2.5	23.2	18.3	18.9
pam50	5	16.8	7.6	20.9	4.8	20.9	1.0	5.6	5.6	4.8	4.7	4.7	20.4	4.8	6.0
ssp2006	5	10.9	10.6	11.1	8.4	13.9	1.5	7.6	5.9	4.5	5.8	4.9	22.7	6.4	8.6
ssp2003	5	12.1	9.5	15.5	7.3	18.8	1.8	7.9	4.6	4.6	4.4	5.4	24.0	3.9	6.0
Average		22.3	16.2	23.2	9.1	25.4	1.5	6.7	10.6	4.0	4.7	4.3	22.9	10.5	11.6