Skip to main content

Table 6 Performance of two Ab-initio methods vs. FIGENIX's structural annotation method

From: FIGENIX: Intelligent automation of genomic annotation: expertise integration in a new software platform

Program

Initial exons(55)

Internal exons(186)

Terminal exons(55)

False positive (overprediction)

Correct full length protein prediction

Genscan

0.55

0.80

0.65

0.22

0.31

HMMGene

0.75

0.81

0.78

0.15

0.38

FIGENIX

0.91

0.92

0.95

0.05

0.87

  1. Performances were measured on a modified version of the HMR195 [53] dataset. The new dataset contains 55 sequences from Mouse and Rat genomes. They were annotated with Genscan, HMMGene and FIGENIX (with the human section of Swissprot [54] as a reference database for homology-based approach).