Markup | Genome Research

Table 3.

Accuracy of Gene Prediction Tools in the Set of Semiartificial Genomic Sequences, When Either Strongly or Moderately Similar Sequences are Used to Model the Genes

Program		Strong similarityP Value < 10⁻⁵⁰ 17 SAG sequences		Moderate similarity 10⁻⁵⁰ <P value < 10⁻⁶ 26 SAG sequences
		Nucleotide		Exon		Nucleotide		Exon
		Nucleotide		Sn		Nucleotide		Sp		$\frac{Sn + Sp}{2}$		Sn	Sp	$\frac{Sn + Sp}{2}$
				Sn				Sp		$\frac{Sn + Sp}{2}$		Sn	Sp	$\frac{Sn + Sp}{2}$
Sn	Sp	CC	Sn	Sp	CC
GenScan	0.91	0.66	0.77	0.67	0.46	0.56	0.91	0.61	0.74	0.67	0.43	0.55
GeneWise	0.99	0.99	0.99	0.90	0.93	0.91	0.68	0.98	0.81	0.46	0.63	0.54
Procrustes	0.92	0.96	0.94	0.80	0.75	0.77	0.66	0.79	0.72	0.48	0.32	0.40

[i] The geometric mean of the P values of the strong similarity sequences was 10⁻¹³⁵ and for the weaker similarity group it was 10⁻³⁹.