The genbank parser fails for certain references

[mihailefter]

When trying to add mRNA and CDS features to their corresponding gene lists, if there is no gene feature present in the record with the same name as found in the sub-features, the parser breaks.

For UD_150167851083 such a case occurs:

gene    <1..13127
        /gene="UGT1A"
        /gene_synonym="GNT1; UGT; UGT1; UGT1A@"
        /db_xref="HGNC:HGNC:12529"

mRNA    join(<6862..6993,7677..7764,8048..8267,12090..13127)
        /gene="UGT1A1"
        /gene_synonym="BILIQTL1; GNT1; HUG-BR1; UDPGT; UDPGT 1-1;
        UGT1; UGT1A"
        /db_xref="HGNC:HGNC:12530"
        /db_xref="MIM:191740"

Most likely the mRNA should be added to the gene, but the /gene qualifier value is different between the two.

[ifokkema]

Note that these are different entities!

UGT1A: HGNC, NCBI

UGT1A1: HGNC, NCBI

[mihailefter]

It seems so from the HGNC also. However, UGT1A is present in the in the mRNA /gene_synonym qualifier.

[ifokkema]

Synonyms change and can get reassigned to other genes. In this case, there is an obvious relationship between the two, but they are not the same.

However, I just saw that UGT1A1 does actually have a /gene tag in UD_150167851083...?

[mihailefter]

Indeed, the UGT1A1 gene is there, but located in the file after the mRNA feature. The bug arises then because the parser tries in one iteration over the record features: (1) to generate the gene, mRNA and CDS objects, and (2) to add the latter (mRNA and CDS) to the previous (gene).

Still, the code would also fail if no UGT1A1 would have been present, so a check whether the gene actually exists should be performed whenever mRNA and CDS objects are added to a gene object.