Table 1.

The majority of the agreement between predictions and microarrays is due to the genome-specific distance models

Genome	Distance	Comparative	All features
E. coli K12	0.406	0.401	0.494
B. subtilis	0.420	0.335	0.461
H. pylori	0.275	0.231	0.343
C. trachomatis	0.260	0.167	0.303
Synechocystis	0.159	0.222	0.268
Halobacterium	0.198	0.159	0.215

For each genome we show the Spearman correlation between the microarray similarity (the Pearson correlation of the normalized log-ratios for two adjacent genes) and the predicted probability that the two genes are in the same operon [P(Operon)] using just intergenic distance, using just the comparative/functional features, or using all features.