Table 1.

Details of the 12 datasets that were analyzed.

ID	G/P	P/D	Number of variables (Original)	Number of variables (PAIFE)	Sample(Class1,Class2)	Outcome variable	Reference
1	G	D	6584	1972	61(40,21)	Colon Cancer	Alon et al [6]
2	G	P	5372	858	86(69,17)	Lung Cancer	Beer et al. [7]
3	P	D	70	15	205(66,139)	Lung Cancer	Bigbee, et al. [5]
4	G	D	7129	2288	72(47,25)	Leukemia	Golub, et al. [8]
5	G	D	7464	1880	36(18,18)	Breast Cancer	Hedenfalk et al. [9]
6	G	P	7129	699	60(20,40)	Hepatocellular carcinoma	Iizuka et al. [10]
7	G	P	7399	1084	240(138,102)	Lymphoma	Rosenwald, et al. [11]
8	G	D	7129	1927	77(58,19)	Lymphoma	Shipp, et al. [12]
9	G	P	24481	4251	78(44,34)	Breast cancer	Van’t Veer, et al. [13]
10	G	D	7039	1230	39(35,4)	Ovarian Cancer	Welch, et al. [14]
11	G	P	12625	1166	249(201,48)	Leukemia	Yeoh, et al. [15]
12	P	D	16	12	583(184,401)	Lung cancer	LungSPORE (unpublished)

G/P indicates if the data is Genomic or Proteomic. P/D shows whether the data is Prognostic (P) or Diagnostic (D). The number of variables (Original) gives the total variables in the original dataset. The number of variables (PAIFE) gives the total number of variables after processing the dataset through our irrelevant feature elimination algorithm ‘PAIFE’. The Sample (Class1, Class2) gives the total number of samples and class distribution, and ‘Reference’, the relevant reference to the dataset.