cuffcompare <outprefix>.stats question

[arrchi]

Hi there,

The values in the outprefix.stats of my cuffcompare are very low. Here is an example:

#cuffcompare -V -r hg19.ensembl-for-tophat.gtf transcripts.gtf
#

#= Summary for dataset: transcripts.gtf :
# Query mRNAs : 65495 in 64983 loci (5211 multi-exon transcripts)
# (421 multi-transcript loci, ~1.0 transcripts per locus)
# Reference mRNAs : 99330 in 43503 loci (82822 multi-exon)
# Corresponding super-loci: 5232
#--------------------| Sn | Sp | fSn | fSp
Base level: 4.6 17.3 - -
Exon level: 0.9 4.3 1.4 6.5
Intron level: 2.5 80.2 2.5 81.7
Intron chain level: 0.1 1.4 0.2 2.6
Transcript level: 0.0 0.0 0.0 0.0
Locus level: 0.2 0.1 0.2 0.2
Missed exons: 333400/353286 ( 94.4%)
Wrong exons: 56158/73680 ( 76.2%)
Missed introns: 262639/272474 ( 96.4%)
Wrong introns: 790/8364 ( 9.4%)
Missed loci: 36101/43503 ( 83.0%)
Wrong loci: 54293/64983 ( 83.5%)

Total union super-loci across all input datasets: 59557

Have anyone seen such low values? Does 94.4% means my data could not map to 94.4% of exons on my reference genome? What does "Wrong exons" mean?

Thanks for any input!
-A