How to fix GTF files by adding specific strings into empty gene_id ""

Question

[sasa]

Hi,

I want to repair GTF file by adding a unique string (such as Product name) to empty gene_id "". I would really appreciate it if anyone could provide any solution.

For example:

grep -m1 'gene_id ""' mygtf.gtf

NC_001717.1 RefSeq  exon    1004    1071    .   +   .   **gene_id ""**; transcript_id "unknown_transcript_1"; anticodon "(pos:1034..1036)"; gbkey "tRNA"; note "putative"; product "tRNA-Phe"; exon_number "1";

I want to add the product name between the double quotes right after the gene_id like:

NC_001717.1 RefSeq  exon    1004    1071    .   +   .   gene_id "tRNA-Phe"; transcript_id "unknown_transcript_1"; anticodon "(pos:1034..1036)"; gbkey "tRNA"; note "putative"; product "tRNA-Phe"; exon_number "1";

I have 24 empty gene_id, and need to fix all of them. I obtained this file from NCBI RefSeq. Unfortunately, this species is not available from the Ensemble database.

The original reason why I would like to fix the GTF file is to filter GTF file with cellragner mkgtf. I am getting the below error, so I need to modify the GTF file.

cellranger.reference.GtfParseError: Error while parsing GTF file /~/genome/mygtf.gtf Property 'gene_id' is empty in GTF line 1809658: NC_001717.1 RefSeq exon 1004 1071 . + gene_id ""; transcript_id "unknown_transcript_1"; anticodon "(pos:1034..1036)"; gbkey "tRNA"; note "putative"; product "tRNA-Phe"; exon_number "1";

Thank you!

Answer 1

[Juke34]

AGAT has a script for that type of task. Look at agat_sp_manage_attributes.pl with --att product/gene_id --cp --overwrite.
I doubt you choose the good solution. Different genes may have the same product value. The best would be to use agat_sp_manage_attributes.pl to remove all gene_id attributes. Then just use agat_convert_sp_gff2gtf.pl it should recreate proper gene_id attributes.

[sasa]

Thank you so much for your suggestion. And, yes, my idea to use the product name was not great. I was able to convert my gtf by using the functions you listed above from AGAT. Thanks again!