| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Gene predictor sensitivity and specificity, averaged for
12 Drosophila species Table P1. Gene predictors compared
for sensitivity and specificity, averaged over all species. Sensitivities for old (Eold) genes are determined by overlap of prediction
exons with homology matches. Sensitivity for new genes (Enew.O and Enew.P)
are determined from overlap with species gene expression exons (O) and
phylogenetic assessed Dmel exons.
Specificity, Pv, is 1 Ð (
predicted exons with none of the above evidence / total predicted exons). Ntr is number of predicted transcripts (distinct genes
for some predictors). Predictor
classes (Ab-initio, Ab-initio with homology guidance, Homology mapping,
Combiner) are discussed in the text.
The random predictor is a randomly sampled null hypothesis for
predictions.
Source:
http://insects.eugenes.org/species/news/genome-summaries/genepredictions-compared.txt,
D.G. Gilbert, Jan. 2007 Grouped
Predictor Key: A =
pure ab-initio, AH= ab-initio with homology evidence, H = homology gene mapper,
C = Combiner of other predictors Group ID PredictorA DL_SNP: DGIL_SNP: SNAP predictor,
no homology A RI_GID: RGUI_GID: geneid
predictor A BZ_CNA: BATZ_CNA: CONTRAST
predictor, no-homology set AH BN_NSC: BREN_NSC: N-SCAN predictor
using homology evidence AH DL_SNO: DGIL_SNO: SNAP predictor using
homology evidence AH NI_GNO: NCBI_GNO: GNOMON predictor
using homology evidence C EE_GLN: EISE_GLN: Glean
prediction combiner C EE_GLR: EISE_GLR: Glean
recombined with higher weight to gene mappers C MD_JIG: UMD_JIG : Jigsaw
prediction combiner H EE_CEX: EISE_CEX: Exonerate (Dmel gene mapping) H EE_CGM: EISE_CGM: GeneMapper
(Dmel gene mapping) H EE_CGW: EISE_CGW: GeneWise (Dmel gene mapping) H OD_GPI: OXFD_GPI: Oxford gene
pipeline (exonerate with Dmel gene mapping) H PH_GMP: PACH_GMP: GeneMapper (Dmel
gene mapping) Predictor
Statistic: Eold = Sensitivity for "Old genes" (HSP_modDM,
Dmel protein matches; about 40,000 exon locations/species) Enew.O = Sensitivity for "New
genes" (OLIV_EXP - HSP_modDM;
5,000 to 10,000 exon locations/species) Enew.P = Sensitivity for "New
genes" (new Dmel exons by phylogenetic analysis; about 2,000 exon
locations/species) Pv = Specificity to exons with evidence
(around 50,000 locations have evidence, and 100,000 lack evidence/species. How
many of the no evidence locations have predictions) Sensitivity to Old/New
genes = ( predicted.overlaps.evidence / total.evidence.exons) Specificity
= 1 - (
no_evidence.predicted.exons / total.predicted.exons ) |