class_code 'u' in all transcripts

[karlaarz]

Hi,

I have two different gtf files (one created from short reads and the other from long reads), and I would like to compare them to see if any of the transcripts are similar among them. I am using the command:

gffcompare gof_filt.gtf gof_transc_in_flair.gtf -o gffcomp_short_flair_gof

However, the results of the.tracking file only have the class_code 'u', and the stats are:

# gffcompare v0.11.2 | Command line was:
# gffcompare gof_filt.gtf gof_transc_in_flair.gtf -o gffcomp_short_flair_gof
#

#= Summary for dataset: gof_filt.gtf 
#     Query mRNAs :    5042 in     216 loci  (5035 multi-exon transcripts)
#            (211 multi-transcript loci, ~23.3 transcripts per locus)

#= Summary for dataset: gof_transc_in_flair.gtf 
#     Query mRNAs :    1033 in     165 loci  (1032 multi-exon transcripts)
#            (129 multi-transcript loci, ~6.3 transcripts per locus)

 Total union super-loci across all input datasets: 381 
  (340 multi-transcript, ~15.9 transcripts per locus)
6075 out of 6075 consensus transcripts written in gffcomp_short_flair_gof.combined.gtf (0 discarded as redundant)

I am not sure what can be wrong, as even the already annotated transcripts are tagged as "unknown" (class_code 'u').

Any help is appreciated!

Thanks

[RobAlbn]

Hi, I am not a GFFCompare developer, but I have just read your issue. In order to compare two GTF files, one of the two GTF files has to be specified as the reference annotation file with the -r parameter. Transcripts of the other file will be compared to transcripts in the reference file, and they will be classified depending on their relationships with reference transcripts (you can find class codes and their meanings on this GFFCompare documentation page in the section "Transcript classification codes": https://ccb.jhu.edu/software/stringtie/gffcompare.shtml). I hope this helps, even if your issue is quite old now.