Table 2.

Experimental sensitivities of the optimal seeds.

seed	mixed	gal	mm1	mmX	pan	average
111**1*11**1*1*111a	0.73837 (1)	0.78102 (2)	0.79375 (2)	0.71692 (1)	0.71909 (3)	0.74983 (1)
111*1**1*1**11*111b	0.71615 (5)	0.77729 (3)	0.76228 (6)	0.69906 (2)	0.69820 (6)	0.73060 (5)
111*1**1*111*111c	0.73103 (2)	0.77070 (4)	0.80484 (1)	0.69386 (3)	0.71802 (4)	0.74369 (2)
1111*111*1111d	0.72246 (4)	0.78234 (1)	0.78821 (3)	0.67856 (5)	0.72535 (1)	0.73939 (3)
1111*111**1*111e	0.72300 (3)	0.76938 (6)	0.78317 (4)	0.68591 (4)	0.72156 (2)	0.73660 (4)
111**1*1**11**1*111f	0.71110 (6)	0.76719 (5)	0.77384 (5)	0.67652 (6)	0.70356 (5)	0.72644 (6)
11111111111g	0.60068 (7)	0.70459 (7)	0.66819 (7)	0.57455 (7)	0.57556 (7)	0.62471 (7)

The experimental sensitivities of the optimal seeds are calculated from the five biological data sets(mixed, gal, mm1, mmX, and pan). The tested seeds are listed as below: a: the optimal seed of HI [0, 1] and HI [0.5, 1], b: the default seed of PatternHunter, c: the optimal seed of HI [0.3, 0.7], d: the optimal seed computed by mandala, e: the seed representing 5th-order non-coding markov model in [5], f: the seed representing 0th-order non-coding markov model in [5], g: the default seed of BLAST.