Table 1.

TIS prediction accuracy of our algorithm (TICO) in comparison with other post-processing tools. RBSfinder [10], GS-Finder [4] and MED-Start [5] were used as post-processors on the same GLIMMER2.02-prediction as TICO. Accuracy was measured in percent of TIS that were predicted correctly with respect to reference annotations. Datasets are explained in detail in section »Results«.

Dataset	total number	%detected ORFs	% correctly predicted TIS
			GLIMMER	MED-Start	TICO	GS-Finder	RBSfinder
EcoGene	854	99.3	63.2	92.0	94.2	90.3	81.9
Link	195	100	66.7	95.4	94.9	92.3	80.0
Bsub	1248	98.6	61.3	89.2	89.4	87.9	78.5
	58	98.3	69.0	94.8	91.4	94.8	82.8
PseudoCAP	3281	97.5	57.8	3.6	85.2	83.6	67.7
R. solanac. chr	3440	97.2	51.5	5.0	74.9	71.4	56.8
R. solanac. plasmid	1676	97.0	48.9	6.0	70.1	66.2	55.5
B. pseudom. chr1	3399	97.7	53.2	5.5	69.6	64.3	53.3
B. pseudom. chr2	2329	97.7	48.9	4.7	67.0	67.5	52.1