[ensembl-dev] Missing gene symbol for VEP with refseq

Wallace Ko myko at l3-bioinfo.com
Tue Jun 7 09:44:19 BST 2016


Hi there,

There is probably a bug on cache in Variation API with annotating with
refseq transcript.

For a single variant A "1:g.121116121T>C", the online VEP (84) result is:
http://grch37.ensembl.org/Homo_sapiens/Tools/VEP/Results?db=core;tl=jIjMm1R03KeDvYW4-1799042
Observe that the 2 rows with Feature type Transcript contains Symbol
SRGAP2C.

When variant A is analysed together with another variant B
"1:g.120935661T>C", the result is:
http://grch37.ensembl.org/Homo_sapiens/Tools/VEP/Results?db=core;tl=dNaMvffRoZ6scMMp-1799049;field1=Location;operator1=is;value1=1:121116121-121116121
Observe that Symbol is missing from second row with Feature type Transcript.

In the subroutine fetch_transcripts of Bio::EnsEMBL::Variation::Utils::VEP (
https://github.com/Ensembl/ensembl-variation/blob/release/84/modules/Bio/EnsEMBL/Variation/Utils/VEP.pm#L3738
):

my %seen_trs;
> ...
> foreach my $chr(...) {
>     foreach my $region(...) {
>         ...
>         my %refseq_stuff = ();
>         if(defined($tmp_cache->{$chr})) {
>             TRANSCRIPT: while(my $tr = shift @{$tmp_cache->{$chr}}) {
>                 ...
>                 if($seen_trs{$dbID}) {
>                     $count_duplicates++;
>                     next;
>                 }
>                 ...
>                 if(defined($config->{refseq}) ||
> defined($config->{merged})) {
>                     # put data to $refseq_stuff
>                 }
>                 $seen_trs{$dbID} = 1;
>                 ...
>             }
>         }
>         ...
>     }
> }


The scope of variable %seen_trs is for all regions is all chromosomes while
the scope of variable %refseq_stuff is for a single region only.
In the second analysis above, transcript for region of variant B is loaded
to cache and marked seen using %seen_trs. When it came to region of variant
A, cache loading is skipped according to %seen_trs but the
%refseq_stuff variable
is actually empty for this new region.

Regards,
Wallace Ko
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ensembl.org/pipermail/dev_ensembl.org/attachments/20160607/9b9de342/attachment.html>


More information about the Dev mailing list