Table 2.

Distribution of the gene-prediction candidates among the four groups that differ by the level of extrinsic support

Species	Support for predicted gene candidates	Order excluded DB		Species excluded DB
		No. of candidates	Precision %	No. of candidates	Precision %
C. elegans	Fully extrinsic	7676	88.9	10,778	91.6
	Partially extrinsic	4804	56.4	5417	54.4
	Match to low-scoring hints	4020	54.7	1548	45.2
	No extrinsic match	1298	24.9	778	18.0
A. thaliana	Fully extrinsic	16,445	97.2	18,083	97.5
	Partially extrinsic	4825	64.4	5807	55.7
	Match to low-scoring hints	1794	50.2	1360	30.1
	No extrinsic match	2964	27.9	1128	9.4
D. melanogaster	Fully extrinsic	8059	95.1	9952	96.8
	Partially extrinsic	2328	49.3	2751	44.9
	Match to low-scoring hints	1043	57.1	165	44.9
	No extrinsic match	1369	41.6	377	15.9
S. lycopersicum	Fully extrinsic	17,639	95.2	18,420	95.0
	Partially extrinsic	5174	47.3	5813	44.3
	Match to low-scoring hints	1577	38.4	1484	29.7
	No extrinsic match	4714	14.8	3703	9.2
D. rerio	Fully extrinsic	15,691	89.8	15,501	92.6
	Partially extrinsic	10,905	16.6	11,769	16.6
	Match to low-scoring hints	1973	11.4	1663	7.3
	No extrinsic match	12,534	0.8	11,879	0.3
G. gallus	Fully extrinsic	11,856	89.3	11,547	89.9
	Partially extrinsic	4857	19.6	5337	20.1
	Match to low-scoring hints	527	8.9	579	7.1
	No extrinsic match	11,332	0.4	11,352	0.3
M. musculus	Fully extrinsic	13,556	94.6	13,769	96.2
	Partially extrinsic	7376	20.6	7606	19.6
	Match to low-scoring hints	957	10.1	1155	7.3
	No extrinsic match	20,711	1.2	19,666	0.5

In the “match to low-scoring hints” category, we placed the candidates that were predicted fully ab initio, whereas in a posteriori analysis, a match of at least one exon border to a low-scoring extrinsic hint was detected. The category “no extrinsic match” contains the ab initio predictions that have no match to any extrinsic hints (see text). Average gene-level Pr values are listed for each group. Protein databases are described in the section “Data sets.”