Convert an 'intron-style' GFF3 file into an 'exon-style' GFF3 file

Question

[Dan]

I have a GFF3 file that doesn't have exons, instead it has introns, UTRs, start and stop codons:

0001.scaffold00002      AUGUSTUS        gene    1386    2772    0.12    +       .       ID=Bv_00001z1_qhas;Name=Bv_00001z1_qhas
0001.scaffold00002      AUGUSTUS        mRNA    1386    2772    0.12    +       .       ID=Bv_00001z1_qhas.t1;Parent=Bv_00001z1_qhas;Name=Bv_00001z1_qhas.t1 0%;Note=cDNAcoverage_0%
0001.scaffold00002      AUGUSTUS        five_prime_UTR  1386    1976    .       +       .       ID=Bv_00001z1_qhas.t1.UTR;Parent=Bv_00001z1_qhas.t1
0001.scaffold00002      AUGUSTUS        start_codon     1977    1979    .       +       0       ID=Bv_00001z1_qhas.t1.start_codon;Parent=Bv_00001z1_qhas.t1
0001.scaffold00002      AUGUSTUS        CDS     1977    2325    0.96    +       0       ID=Bv_00001z1_qhas.t1.CDS;Parent=Bv_00001z1_qhas.t1
0001.scaffold00002      AUGUSTUS        intron  2326    2619    0.81    +       .       ID=Bv_00001z1_qhas.t1.intron;Parent=Bv_00001z1_qhas.t1
0001.scaffold00002      AUGUSTUS        CDS     2620    2747    0.8     +       2       ID=Bv_00001z1_qhas.t1.CDS;Parent=Bv_00001z1_qhas.t1
0001.scaffold00002      AUGUSTUS        stop_codon      2745    2747    .       +       0       ID=Bv_00001z1_qhas.t1.stop_codon;Parent=Bv_00001z1_qhas.t1
0001.scaffold00002      AUGUSTUS        three_prime_UTR 2748    2772    .       +       .       ID=Bv_00001z1_qhas.t1.UTR;Parent=Bv_00001z1_qhas.t1

I can convert this to 'exon-style' by calculating the exons from the above, but I'm wondering if there is an 'off the shelf' solution?

Cheers,
Dan.

Answer 1

[Dan]

Actually, this can be done with GenomeTools. The dupfeat command duplicates features of type -source and outputs the copies with type dest. The mergefeat command merges adjacent features of the same type:

gt dupfeat -dest exon -source CDS your.gff3 \
  | gt dupfeat -dest exon -source three_prime_UTR \
  | gt dupfeat -dest exon -source five_prime_UTR \
  | gt mergefeat \
  | gt gff3 -retainids -sort -tidy -o your.new.gff3

Pretty slick!

[Istvan Albert]

good to know

[Daniel Standage]

GenomeTools is the answer for many questions I have about GFF3 processing!

Answer 2

[Dan]

Here is my answer in full, complicated by the fact that the dumb format wasn't consistent in it's stupidity:

Answer 3

[Istvan Albert]

Looks like your CDS' are the exons, only that the CDS' also include the stop codon that is not actually part of the mRNA.

I don't think that there is a tool to do what you need in one step.

[Dan]

Right, the CDS are the exons except when interrupted by a start (stop) codon, in which case the exon includes the five (three) prime UTR.... I guess?

[Istvan Albert]

the definition for these is actually a lot more complicated, and I suspect tool developers may be a little cavalier in labeling. I would not be surprised if there were inconsistencies along the way. It all depends what is the file needed for.

Exon: http://www.sequenceontology.org/browser/current_svn/term/SO:0000147

CDS: http://www.sequenceontology.org/browser/current_svn/term/SO:0000316

[Dan]

The definitions are (now) clear (and the GFF validates OK), the pain is knowing if your CDS abuts a five (three) prime UTR (or both!) and if your five (three) prime UTR is a separate exon... Actually, my solution has been ignoring the intron features, these let me solve it actually! I'll post Perl when I'm done.