Address and Filter NCBI Gene IDs misassigned due to read-through transcripts

[ACastanza]

There seems to be an issue which appears to originate on the NCBI side, whereby genes with a read-through transcript can end up getting the NCBI gene ID of the read through assigned to one of(?) the parent Ensembl Genes.

Here's an example from Biomart (taken in Ensembl 103) which demonstrates this issue:

Ensembl Gene ID	NCBI Gene ID	HGNC Gene ID	Gene Symbol	Gene Title
ENSG00000278232	1394	HGNC:2357	CRHR1	corticotropin releasing hormone receptor 1 [Source:HGNC Symbol;Acc:HGNC:2357]
ENSG00000278232	104909134	HGNC:51483	CRHR1	corticotropin releasing hormone receptor 1 [Source:HGNC Symbol;Acc:HGNC:2357]
ENSG00000282456	104909134	HGNC:51483	LINC02210-CRHR1	LINC02210-CRHR1 readthrough [Source:HGNC Symbol;Acc:HGNC:51483]
ENSG00000204650	147081	HGNC:26327	LINC02210	long intergenic non-protein coding RNA 2210 [Source:HGNC Symbol;Acc:HGNC:26327]

https://www.ncbi.nlm.nih.gov/gene/?term=1394
https://www.ncbi.nlm.nih.gov/gene/?term=104909134

From a quick look at your genes sheet:

ensembl_gene_id	ensembl_gene_version	gene_symbol	gene_symbol_source_db	gene_symbol_source	gene_biotype	gene_description
ENSG00000278232	4	LINC02210-CRHR1	HGNC	HGNC:51483	protein_coding	LINC02210-CRHR1 readthrough [Source:HGNC Symbol;Acc:HGNC:51483]
ENSG00000282456	1	LINC02210-CRHR1	HGNC	HGNC:51483	lncRNA	LINC02210-CRHR1 readthrough [Source:HGNC Symbol;Acc:HGNC:51483]

it would seem to be affected

[dhimmel]

Here's an example where ENSG00000004866 (ST7) maps to both ST7 and ST7-OT3 in ncbigene:

ensembl_representative_gene_id	ensembl_gene_id	gene_symbol	xref_source	xref_accession	xref_label	xref_description	xref_info_type	xref_linkage_annotation	xref_curie
ENSG00000004866	ENSG00000004866	ST7	EntrezGene	7982	ST7	suppression of tumorigenicity 7	DEPENDENT	None	ncbigene:7982
ENSG00000004866	ENSG00000004866	ST7	EntrezGene	93655	ST7-OT3	ST7 overlapping transcript 3	DEPENDENT	None	ncbigene:93655

I don't think "ST7 overlapping transcript 3" is a read-through, although perhaps this is a similar situation where we don't want to be mapping to both ncbigenes.

[dhimmel]

Here's an example of a read-through:

ensembl_representative_gene_id	ensembl_gene_id	gene_symbol	xref_source	xref_accession	xref_label	xref_description	xref_info_type	xref_linkage_annotation	xref_curie
ENSG00000088298	ENSG00000088298	EDEM2	EntrezGene	111089941	MMP24-AS1-EDEM2	MMP24-AS1-EDEM2 readthrough	DEPENDENT	None	ncbigene:111089941
ENSG00000088298	ENSG00000088298	EDEM2	EntrezGene	55741	EDEM2	ER degradation enhancing alpha-mannosidase lik...	DEPENDENT	None	ncbigene:55741

What I'm thinking is that for a given representative ensembl gene, we can pick the NCBI gene mapping with the same symbol. That would address these issues as well as #10.