Gene Stats Detail


Summary

  count total len % cov % gc min len max len median len mean len
Contig 202 11525919 99.73 37.67 504 778483 3512.00 57059.00
Gene 5155 10377388 83.16 38.54 153 15146 1764.00 2013.07
Exon 11129 9972157 79.29 38.81 3 14434 457.00 896.05
Coding 5180 7031070 60.61 40.68 135 14877 1104.00 1357.35
Intergenic 3599 1946324 16.84 33.31 1 36800 297.00 540.80
5' UTR 4180 1206456 9.93 34.50 1 4602 173.00 288.63
3' UTR 4219 1734631 13.90 33.86 3 4776 254.00 411.15

Specificity/Sensitivity Analysis

close

Terminology is from Burset M and Guigo R, "Evaluation of gene structure prediction programs." Genomics, 1996 Jun 15;34(3):353-67.
TP (true positives): nucleotides predicted as exonic in predictions and EST evidence; splice junctions with exact agreement, in position and type (donor:donor, acceptor:acceptor), between predictions and EST evidence; exons with both splice sites in exact agreement with EST evidence.
TN (true negatives): nucleotides predicted as intronic in predictions and EST evidence; not defined for splice sites or exons.
FP (false positives/overpredictions): nucleotides predicted as exonic where EST evidence indicates an intron; predicted splice junctions where the EST alignments do not have a splice*; predicted exons which fall entirely within an intron supported by EST evidence ("wrong" in Guigo's terminology).
FN (false negatives/underpredictions): nucleotides predicted as intronic where EST evidence indicates an exon; EST alignment splice junctions where the predictions do not have a splice*; EST-based exons that fall entirely within a predicted intron ("miss" in Guigo's terminology).
unknown (cannot be scored): nucleotides that do not touch an EST alignment; splices that do not touch an EST alignment, and initial/terminal splices that fall within an initial/terminal exon of a partial EST alignment; exons that either do not touch an EST alignment or fully encapsulate a single-exon partial EST alignment.
sn (sensitivity): TP/(TP+FN)
sp (specificity): TP/(TP+FP)
cov (coverage): percentage of predicted elements that are not "unknown"

In the very rare cases where the query predicts a donor site exactly where EST evidence indicates an acceptor (or vice versa), we count it as both FP and FN. Certain classes of incorrect exon predictions are not labeled TP, FP nor FN.

Compared 5180 predictions to 4859 reference models.

  cov sn sp TP TN FP FN unknown
nucleotide 0.9581 0.9960 0.9996 9571021 373770 3928 38818 436477
splices 0.8129 0.9927 0.9463 17122 - 972 126 4164
exons 0.9034 0.9359 0.9045 9094 - 4 3 1075