Wrong naming of transcript in gtf/gff3

Question

[Diego]

Hi all,

Can someone tell me why naming transcript_id "jg1.t1"; gene_id "jg1"; at a gtf is not proper GFF file?

Thanks in advance

[Pierre Lindenbaum]

who said it's wrong ?

[Diego]

Hi,

Softwares/packages such as EvidenceModeler, Augustus, and gff3tolkit always says that the ID is wrong.

CAKLNU010000942.1   other_pred1 CDS 724 1083    0.38    +   0   transcript_id "jg1.t1"; gene_id "jg1";

[Pierre Lindenbaum]

This is a CDS, is there a transcript with transcript_id "jg1.t1" and/or a gene with gene_id "jg1" in the very same file ?

[Diego]

yes. The gtf look like this

CAEGAH010020538.1      AUGUSTUS       transcript     407    7407   0.08   -      .      transcript_id "g1.t1"; gene_id "g1";
CAEGAH010020538.1      AUGUSTUS       exon   407    547    .      -      .      transcript_id "g1.t1"; gene_id "g1";
CAEGAH010020538.1      AUGUSTUS       exon   6202   7407   .      -      .      transcript_id "g1.t1"; gene_id "g1";
CAEGAH010020538.1      AUGUSTUS       CDS    407    547    .      -      0      transcript_id "g1.t1"; gene_id "g1";
CAEGAH010020538.1      AUGUSTUS       CDS    6202   7407   .      -      0      transcript_id "g1.t1"; gene_id "g1";

[Diego]

or like this one I have converted it into gff3:

CAKLNU010000942.1       gffcl   locus   724     2835    .       +       .       ID=RLOC_00000001;transcripts=jg1.t1
CAKLNU010000942.1       AUGUSTUS        transcript      724     2835    .       +       .       ID=jg1.t1;geneID=jg1;locus=RLOC_00000001
CAKLNU010000942.1       AUGUSTUS        CDS     724     1083    .       +       0       Parent=jg1.t1
CAKLNU010000942.1       AUGUSTUS        CDS     1181    1625    0.34    +       0       Parent=jg1.t1
CAKLNU010000942.1       AUGUSTUS        CDS     2270    2835    0.42    +       2       Parent=jg1.t1

Answer 1

[Juke34]

GFF formats require ID attributes for features (gene, transcript, exon, etc.) while in GTF you must have gene_id for all features and ?transcript_id for all features excepted gene feature. For more details of differences between the two format, you can read this: https://agat.readthedocs.io/en/latest/gxf.html

[Diego]

I think this is what I looking for: agat_convert_sp_gxf2gxf.pl

[Diego]

thanks for the very useful reply. I wonder why most tools are gff2gtf and not in the other way around. I guess these tools can do gft2gff as well. I am going to check agat

[Diego]

can agat_sp_ensembl_output_style.pl do this: -g non-ensembl.gtf [ -o ensembl_like.gtf ] ?

[Juke34]

Potentially yes but you have to use AGAT >=1.0.0 and modify the AGAT config to specify that you want GTF output.