Automatic annotation of eukaryotic genes, pseudogenes and promoters

Table 5 Accuracy of human gene prediction using similar mouse proteins

	Sn ex	Sno ex	Sp ex	Sn nuc	Sp nuc	CC	%CG
Similarity of mouse protein >90% in 921 sequences
Fgenesh	86.2	91.7	88.6	93.9	93.4	0.9334	34
Genwise	93.9	97.6	95.9	99.0	99.6	0.9926	66
Fgenesh+	97.3	98.9	98.0	99.1	99.6	0.9936	81
Prot_map	95.9	98.3	96.9	99.1	99.5	0.9924	73
80% < similarity of mouse protein <90% in 1,441 sequences
Fgenesh	85.8	92.1	87.7	94.0	93.4	0.9334	30
Genewise	92.6	98.0	94.1	98.9	99.5	0.9912	58
Fgenesh+	96.8	99.0	97.2	99.1	99.5	0.9929	77
Prot_map	93.9	98.5	94.1	98.9	99.3	0.9898	60

Sn_ex, sensitivity on the exon level (exact exon predictions); Sno_ex, sensitivity with exon overlap; Sp_ex, specificity on the exon level; Sn_nuc, sensitivity, nucleotides; Sp_nuc, specificity, nucleotides; CC, correlation coefficient; %CG, percent of genes predicted completely correctly (no missing and no extra exons, and all exon boundaries are predicted exactly correctly).

ISSN: 1474-760X