Does table_annovar.pl -vcfinput intentionally change character encoding?

[tkoomar]

I do not know if this is a feature, bug, or user error. When providing a VCF to ANNOVAR via table_annovar.pl -vcfinput the output VCF has altered character encoding for punctuation that is part of some annotations -- i.e. ";" becomes "\x3b" and "=" becomes "\x3d" (illustrated by excerpted portions of multianno.txt and .vcf files below)

From the VCF specification, these characters are reserved as delimiters and should not appear within individual INFO fields. So, while I can see that this behavior may be intentional, I was unable to determine from documentation if this is indeed the case.

multianno.txt

Chr     Start   End     Ref     Alt     Func.refGene    Gene.refGene    GeneDetail.refGene
1       10469   10469   C       0       intergenic      NONE;DDX11L1    dist=NONE;dist=1405
1       10469   10469   C       G       intergenic      NONE;DDX11L1    dist=NONE;dist=1405 

multianno.vcf

#CHROM  POS     ID      REF     ALT     QUAL    FILTER  INFO 
1       10469   rs370233998     C       *       2026.12 GATKCutoffSNP   ANNOVAR_DATE=2018-04-16;Func.refGene=intergenic;Gene.refGene=NONE\x3bDDX11L1;GeneDetail.refGene=dist\x3dNONE\x3bdist\x3d1405
1       10469   rs370233998     C       G       2026.12 GATKCutoffSNP   ANNOVAR_DATE=2018-04-16;Func.refGene=intergenic;Gene.refGene=NONE\x3bDDX11L1;GeneDetail.refGene=dist\x3dNONE\x3bdist\x3d1405

[kaichop]

Yes it is intentional, to differentiate the regular = notation in VCF specification and the "=" within a specific field.

…

On Fri, Sep 7, 2018 at 4:56 PM, tkoomar ***@***.***> wrote: I do not know if this is a feature, bug, or user error. When providing a VCF to ANNOVAR via table_annovar.pl -vcfinput the output VCF has altered character encoding for punctuation that is part of some annotations -- i.e. ";" becomes "\x3b" and "=" becomes "\x3d" (illustrated by excerpted portions of multianno.txt and .vcf files below) From the VCF specification, these characters are reserved as delimiters and should not appear within individual INFO fields. So, while I can see that this behavior may be intentional, I was unable to determine from documentation if this is indeed the case. *multianno.txt* Chr Start End Ref Alt Func.refGene Gene.refGene GeneDetail.refGene 1 10469 10469 C 0 intergenic NONE;DDX11L1 dist=NONE;dist=1405 1 10469 10469 C G intergenic NONE;DDX11L1 dist=NONE;dist=1405 *multianno.vcf* #CHROM POS ID REF ALT QUAL FILTER INFO 1 10469 rs370233998 C * 2026.12 GATKCutoffSNP ANNOVAR_DATE=2018-04-16;Func.refGene=intergenic;Gene.refGene=NONE\x3bDDX11L1;GeneDetail.refGene=dist\x3dNONE\x3bdist\x3d1405 1 10469 rs370233998 C G 2026.12 GATKCutoffSNP ANNOVAR_DATE=2018-04-16;Func.refGene=intergenic;Gene.refGene=NONE\x3bDDX11L1;GeneDetail.refGene=dist\x3dNONE\x3bdist\x3d1405 — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#41>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AFptuLVcoIxAibZDRzL4cgARxaX1GexLks5uYt2RgaJpZM4Wfd2M> .

[tkoomar]

Excellent, thank you for the clarification.

A statement of this (perhaps in docs/articles/VCF.md) might be a nice heads up to new users like myself, but any issue one might have with this behavior can be solved easily enough using sed before bgzipping the resulting VCF. Closing this issue.