Table 1.

Correlations between median expression distance and regulatory similarity. The performance of different distance measures were compared in each of seven mouse experiments: Su et al. (Su), Storch et al. (Circadian), the neocortex development (Cortex), the murine model of human asthma (Lung), the hippocampus samples from neurofibromin-1 heterozygous study (NF), Zhao et al. (Muscle), and Wang et al.(PI). The number of microarrays used in each data set are shown in the first row. The p-values in the parentheses are obtained by permuting the mapping between genes and their promoter regions 500 times.

Expression distance measure	Su 89	Circadian 24	Cortex 17	Lung 39	NF 30	Muscle 54	PI 35
1 – correlation	-0.165 (0.422)	-0.261 (0.158)	-0.636 (0.008)	-0.423 (0.124)	-0.826 (0.000)	-0.358 (0.170)	-0.384 (0.198)
1 – cosine correlation	-0.802 (0.000)	-0.392 (0.106)	-0.679 (0.002)	-0.456 (0.066)	-0.878 (0.000)	-0.047 (0.488)	0.016 (0.666)
Square root 1 – correlation	-0.177 (0.416)	-0.280 (0.230)	-0.636 (0.008)	-0.412 (0.162)	-0.836 (0.000)	-0.401 (0.148)	-0.396 (0.200)
Square root 1 – cosine correlation	-0.783 (0.000)	-0.401 (0.166)	-0.683 (0.002)	-0.464 (0.064)	-0.869 (0.000)	-0.104 (0.490)	-0.007 (0.670)
1 – correlation after log2 transformation	-0.178 (0.310)	-0.254 (0.124)	-0.459 (0.032)	-0.534 (0.030)	-0.798 (0.000)	-0.136 (0.314)	-0.035 (0.428)
1 – cosine correlation after log2 transformation	-0.685 (0.006)	-0.026 (0.346)	-0.833 (0.000)	-0.830 (0.000)	-0.874 (0.000)	-0.540 (0.032)	0.346 (0.812)
The new distance measure	-0.777 (0.000)	-0.789 (0.000)	-0.735 (0.002)	-0.776 (0.000)	-0.859 (0.000)	-0.774 (0.000)	-0.656 (0.000)