Can't get children from NCBI GFF (may be user error)

Hi- I’ve been trying to use gffutils to parse an NCBI GFF and ran into an issue getting the children. I’m testing with a small gff of three genes pulled from the original gff. Things started out OK:

if not test_db:
    db=gffutils.create_db(test, test_db, id_spec={'gene': 'db_xref'})
db=gffutils.FeatureDB(test_db)
for i in db.featuretypes():
    print("Feature: %s: %d" % (i, db.count_features_of_type(i)))

Feature: CDS: 38
Feature: exon: 41
Feature: gene: 3
Feature: mRNA: 6

And, i can get genes

for g in db.features_of_type('gene'):
    print(g)

chr1	BestRefSeq%2CGnomon	gene	11686635	11725857	.	+	.	ID=gene-DRAXIN;Dbxref=GeneID:374946,HGNC:HGNC:25054,MIM:612682;Name=DRAXIN;description=dorsal inhibitory axon guidance protein;gbkey=Gene;gene=DRAXIN;gene_biotype=protein_coding;gene_synonym=AGPA3119,C1orf187,neucrin,UNQ3119
chr1	BestRefSeq	gene	15617458	15669044	.	+	.	ID=gene-DDI2;Dbxref=GeneID:84301,HGNC:HGNC:24578;Name=DDI2;description=DNA damage inducible 1 homolog 2;gbkey=Gene;gene=DDI2;gene_biotype=protein_coding
chr1	BestRefSeq	gene	19920009	19923617	.	-	.	ID=gene-PLA2G2E;Dbxref=GeneID:30814,HGNC:HGNC:13414,MIM:618320;Name=PLA2G2E;description=phospholipase A2 group IIE;gbkey=Gene;gene=PLA2G2E;gene_biotype=protein_coding;gene_synonym=GIIE sPLA2,sPLA2-IIE

and a specific gene

gene=db['gene_1']
print(gene)

chr1	BestRefSeq%2CGnomon	gene	11686635	11725857	.	+	.	ID=gene-DRAXIN;Dbxref=GeneID:374946,HGNC:HGNC:25054,MIM:612682;Name=DRAXIN;description=dorsal inhibitory axon guidance protein;gbkey=Gene;gene=DRAXIN;gene_biotype=protein_coding;gene_synonym=AGPA3119,C1orf187,neucrin,UNQ3119

but no children:

for i in db.children(gene):
    print(i)

Is this just user error or something with the NCBI GFF?

[lldelisle]

I never met a gtf where it did not work but I don’t know… I am using gtf files from ensembl.

[lldelisle]

No all these features can be deduced from exons. In each exon feature you have the transcript_id … if you ask gffutils to build the db it will reconstruct a transcript where the start is the start of the first exon and the end the end of the last exon. You are not loosing any information… You can try and confirm.