Table 1.

Biomedical datasets used for the comparison experiments

#	T	#C	#A	#S	M	Reference
1	D	2	6584	61	0.651	Alon et al. (1999)
2	D	3	12 582	72	0.387	Armstrong et al. (2002)
3	P	2	5372	86	0.795	Beer et al. (2002)
4	D	5	12 600	203	0.657	Bhattacharjee et al. (2001)
5	P	2	5372	69	0.746	Bhattacharjee et al. (2001)
6	D	2	7129	72	0.650	Golub et al., 1999
7	D	2	7464	36	0.500	Hedenfalk et al. (2001)
8	P	2	7129	60	0.661	Iizuka et al. (2003)
9	D	4	2308	83	0.345	Khan et al. (2001)
10	D	4	12 625	50	0.296	Nutt et al. (2003)
11	D	5	7129	90	0.642	Pomeroy et al. (2002)
12	P	2	7129	60	0.645	Pomeroy et al. (2002)
13	D	26	16 063	280	0.574	Ramaswamy et al. (2001)
14	P	2	7399	240	0.145	Rosenwald et al. (2002)
15	D	9	7129	60	0.506	Staunton et al. (2001)
16	D	2	7129	77	0.746	Shipp et al. (2002)
17	D	2	10 510	102	0.150	Singh et al. (2002)
18	D	11	12 533	174	0.150	Su et al. (2001)
19	P	2	24 481	78	0.562	van't Veer et al. (2002)
20	D	2	7039	39	0.878	Welsh et al. (2001)
21	P	2	12 625	249	0.805	Yeoh et al. (2002)
22	D	2	11 003	322	0.784	Petricoin et al. (2002)
23	D	3	11 170	159	0.364	Pusztai et al. (2004)
24	D	2	36 778	52	0.556	Ranganathan (2005)

In the type (T) column, P signifies prognostic and D signifies diagnostic. #C represents the number of classes, #A the number of attributes within the dataset, #S the number of samples and M is the fraction of the data covered by the most frequent target value. The first 21 datasets contain genomic data, whereas the last three datasets contain proteomic data.