'' instead of None for inexistent VEP values

[jdelafon]

Empty values when reading from custom fields added through VEP (such as HGVSc and HGVSp) are represented by the empty string in the database. I think it could be safely replaced by NULL.

sqlite> select vep_hgvsc,vep_hgvsp from variants where vep_hgvsp='' limit 10;
|
|
|
...
sqlite> select vep_hgvsc,vep_hgvsp from variants where vep_hgvsp is NULL;
sqlite>

[jxchong]

Agreed, although currently GEMINI doesn't officially support the most recent versions of VEP that do HGVSc and HGVSp. GEMINI officially only supports up to VEP v75.

@brentp Whenever you get around to officially supporting VEP v83, this is an issue for all custom VEP fields that I've tested in VEP v82 and v83.

[brentp]

ok. this is news to me. any time you find something like this, an example VCF with no genotypes will be a great help. It will likely take me longer to update vep and annotate an example than it will to fix the bug-- so, yeah examples greatly appreciated.

[jxchong]

VEP vcf made with VEP v82/v83

$ gemini load -v test.VEP.nogeno.vcf.gz -t VEP --cores 4 test.VEP.db

test.VEP.nogeno.vcf.gz

$ gemini query -q "SELECT distinct(vep_gene_pheno) from variants" test.VEP.db
$ gemini query -q "SELECT vep_hgvsc,vep_hgvsp from variants where vep_hgvsp='' limit 10" test.VEP.db

$ gemini query -q "SELECT distinct(vep_gene_pheno) from variants" test.VEP.db

None
1   
$ gemini query -q "SELECT vep_hgvsc,vep_hgvsp from variants where vep_hgvsp='' limit 10" test.VEP.db




ENST00000327044.6:c.26+22C>T
ENST00000379409.2:c.865-10A>C
ENST00000379409.2:c.865-6A>C
ENST00000491024.1:c.397+24A>G
ENST00000379389.4:c.-33T>C
ENST00000379370.2:c.5651+5C>T
ENST00000354700.5:c.1407+4C>T
$ gemini query -q "SELECT distinct(exon) from variants limit 10" test.VEP.db
1/1

None
10/14
13/14
16/19
10/19
9/19
4/12
12/12

[brentp]

@jxchong and @muraveill so, I made some changes and now have:

$ gemini query -q "SELECT distinct(vep_gene_pheno) from variants" test.VEP.db
None
1
$ gemini query -q "SELECT vep_hgvsc,vep_hgvsp from variants where vep_hgvsp='' limit 10" test.VEP.db
$ gemini query -q "SELECT vep_hgvsc,vep_hgvsp from variants where vep_hgvsp is NULL limit 10" test.VEP.db
None    None
None    None
None    None
None    None
ENST00000327044.6:c.26+22C>T    None
ENST00000379409.2:c.865-10A>C   None
ENST00000379409.2:c.865-6A>C    None
ENST00000491024.1:c.397+24A>G   None
ENST00000379389.4:c.-33T>C  None
ENST00000379370.2:c.5651+5C>T   None

Does that look sensible? I'm

[jxchong]

Works for me

[brentp]

Thanks guys, this will be in 0.18.1