Variant not being annotated

[kjwinfield]

Describe the issue

When running VEP using a local implementation via Docker, a variant is not being annotated. The variant in question is NM_174936.4:c.63_65dup, represented in GRCh37 VCF format as:

1	55505552	rs113330492	A	ACTG

This is an insertion of CTG in a CTG repeat version and the left-alignment means that it is represented as 1:55505552:A:ACTG, but it could also be represented as 1:55505575:T:TGCT.
We are using a custom annotation via VCF to flag this variant to be excluded, is there a way we can represent it in VCF format, that would cause it to be annotated correctly?

Additional information

This variant is benign but is passing our filtering for possible pathogenic variants as it is not being annotated by VEP, so cannot be filtered out based on annotations showing its ClinVar significance, gnomAD frequency etc.

System

• VEP version: v110.1
• VEP Cache version: 110
• Perl version: N/A - whatever is present in docker container
• OS: 20.04
• tabix installed ? yes

Full VEP command line

docker run -v /home/dnanexus:/data -w /data 199b8c2aa90b vep -i /data/134392201-24352R0101-25NGCEN17-9527-M-99347387_markdup_recalibrated_Haplotyper_temp.vcf -o /data/134392201-24352R0101-25NGCEN17-9527-M-99347387_markdup_recalibrated_Haplotyper_temp_annotated.vcf.gz --dir /data --vcf --cache --refseq --exclude_predicted --symbol --hgvs --hgvsg --check_existing --variant_class --numbers --format vcf --offline --exclude_null_alleles --assembly GRCh37 --custom /data/clinvar_20250415_GRCh37.vcf.gz,ClinVar,vcf,exact,0,CLNSIG,CLNREVSTAT,CLNDN,CLNSIGCONF --custom /data/gnomad.genomes.r2.1.1.sites.all.noVEP_normalised_decomposed_PASS.dias_trimmed_v1.0.0.vcf.bgz,gnomADg,vcf,exact,0,AC,AN,AF,nhomalt,popmax,AC_popmax,AN_popmax,AF_popmax,nhomalt_popmax --custom /data/gnomad.exomes.r2.1.1.sites.noVEP_normalised_decomposed_PASS.dias_trimmed_v1.0.0.vcf.bgz,gnomADe,vcf,exact,0,AC,AN,AF,nhomalt,popmax,AC_popmax,AN_popmax,AF_popmax,nhomalt_popmax,non_cancer_AC,non_cancer_AN,non_cancer_AF,non_cancer_nhomalt,non_cancer_AC_popmax,non_cancer_AN_popmax,non_cancer_AF_popmax,non_cancer_nhomalt_popmax,non_cancer_popmax --custom /data/TWE_POPAF_N500_chr1-22_220413.vcf.gz,TWE,vcf,exact,0,AF,AC_Hom,AC_Het,AN --custom /data/HGMD_Pro_2025.1_hg19.vcf.gz,HGMD,vcf,exact,0,PHEN,RANKSCORE,CLASS --custom /data/GRCh37_optimised_filtering_excluded_variants_v1.0.1.vcf.gz,ExcludeVariant,vcf,exact,0,ExcludeList --custom /data/GRCh37_inclusion_list_v1.0.1.vcf.gz,IncludeVariant,vcf,exact,0,IncludeList --plugin SpliceAI,snv=/data/spliceai_scores.masked.snv.hg19.vcf.gz,indel=/data/spliceai_scores.masked.indel.hg19.vcf.gz --plugin REVEL,/data/revel_b37.tsv.gz --plugin CADD,/data/cadd_whole_genome_SNVs_GRCh37.tsv.gz,/data/gnomad.genomes.r2.1.1.indel.tsv.gz,/data/InDels_GRCh37.tsv.gz --fields Allele,SYMBOL,HGNC_ID,VARIANT_CLASS,Consequence,IMPACT,EXON,INTRON,Feature,HGVSc,HGVSp,HGVS_OFFSET,Existing_variation,STRAND,ClinVar,ClinVar_CLNSIG,ClinVar_CLNSIGCONF,ClinVar_CLNDN,gnomADg_AC,gnomADg_AN,gnomADg_AF,gnomADg_nhomalt,gnomADg_popmax,gnomADg_AC_popmax,gnomADg_AN_popmax,gnomADg_AF_popmax,gnomADg_nhomalt_popmax,gnomADe_AC,gnomADe_AN,gnomADe_AF,gnomADe_nhomalt,gnomADe_popmax,gnomADe_AC_popmax,gnomADe_AN_popmax,gnomADe_AF_popmax,gnomADe_nhomalt_popmax,gnomADe_non_cancer_AC,gnomADe_non_cancer_AN,gnomADe_non_cancer_AF,gnomADe_non_cancer_nhomalt,gnomADe_non_cancer_AC_popmax,gnomADe_non_cancer_AN_popmax,gnomADe_non_cancer_AF_popmax,gnomADe_non_cancer_nhomalt_popmax,gnomADe_non_cancer_popmax,TWE_AF,TWE_AC_Hom,TWE_AC_Het,TWE_AN,HGMD,HGMD_PHEN,HGMD_CLASS,HGMD_RANKSCORE,ExcludeVariant_ExcludeList,IncludeVariant_IncludeList,SpliceAI_pred_DS_AG,SpliceAI_pred_DS_AL,SpliceAI_pred_DS_DG,SpliceAI_pred_DS_DL,SpliceAI_pred_DP_AG,SpliceAI_pred_DP_AL,SpliceAI_pred_DP_DG,SpliceAI_pred_DP_DL,REVEL,CADD_PHRED --buffer_size 500 --fork 16 --no_stats --compress_output bgzip --shift_3prime 1

[ainefairbrother]

Hi @kjwinfield,

As I understand, this variant is on GRCh38, not 37. Currently your VEP command runs using --assembly GRCh37 and a GRCh37 cache. Could you check whether adjusting this means you're able to annotate?

Let me know whether this works or you have any further issues.

Best,

@ainefairbrother

[kjwinfield]

hi @ainefairbrother this is a GRCh37 variant and the VCF notation is for GRCh37, see below screenshot from VariantValidator

Best wishes,
@kjwinfield

[ainefairbrother]

Hi @kjwinfield,

OK, thanks for confirming that - you're correct. I will try to reproduce this issue and get back to you. Could I see the exact output that you're getting for this variant?

Best,

@ainefairbrother

[kjwinfield]

hi @ainefairbrother,

Thank you for looking into this. The output in VCF is this:

1	55505552	rs113330492	A	ACTG	23917.8	.	AC=1;AF=0.5;AN=2;BaseQRankSum=-1.118;ClippingRankSum=0;DB;DP=2451;ExcessHet=3.0103;FS=1.09;MLEAC=1;MLEAF=0.5;MQ=59.98;MQRankSum=0;QD=12.13;ReadPosRankSum=1.004;SOR=0.627;CSQ=GCT|PCSK9||insertion|inframe_insertion|MODERATE|1/12||NM_174936.4|NM_174936.4:c.63_65dup|NP_777596.2:p.Leu22_Leu23insCys|23||1|||||||||||||||||||||||||||||||||||||||||||||||||||13.35	GT:AD:DP:GQ:PL	0/1:1007,965:2026:99:23946,0,38815

and the expected fields are:

##VEP-command-line='vep -i /data/134392201-24352R0101-25NGCEN17-9527-M-99347387_markdup_recalibrated_Haplotyper_temp.vcf -o /data/134392201-24352R0101-25NGCEN17-9527-M-99347387_markdup_recalibrated_Haplotyper_temp_annotated.vcf.gz --dir /data --vcf --cache --refseq --exclude_predicted --symbol --hgvs --hgvsg --check_existing --variant_class --numbers --format vcf --offline --exclude_null_alleles --assembly GRCh37 --custom /data/clinvar_20250415_GRCh37.vcf.gz,ClinVar,vcf,exact,0,CLNSIG,CLNREVSTAT,CLNDN,CLNSIGCONF --custom /data/gnomad.genomes.r2.1.1.sites.all.noVEP_normalised_decomposed_PASS.dias_trimmed_v1.0.0.vcf.bgz,gnomADg,vcf,exact,0,AC,AN,AF,nhomalt,popmax,AC_popmax,AN_popmax,AF_popmax,nhomalt_popmax --custom /data/gnomad.exomes.r2.1.1.sites.noVEP_normalised_decomposed_PASS.dias_trimmed_v1.0.0.vcf.bgz,gnomADe,vcf,exact,0,AC,AN,AF,nhomalt,popmax,AC_popmax,AN_popmax,AF_popmax,nhomalt_popmax,non_cancer_AC,non_cancer_AN,non_cancer_AF,non_cancer_nhomalt,non_cancer_AC_popmax,non_cancer_AN_popmax,non_cancer_AF_popmax,non_cancer_nhomalt_popmax,non_cancer_popmax --custom /data/TWE_POPAF_N500_chr1-22_220413.vcf.gz,TWE,vcf,exact,0,AF,AC_Hom,AC_Het,AN --custom /data/HGMD_Pro_2025.1_hg19.vcf.gz,HGMD,vcf,exact,0,PHEN,RANKSCORE,CLASS --custom /data/GRCh37_optimised_filtering_excluded_variants_v1.0.1.vcf.gz,ExcludeVariant,vcf,exact,0,ExcludeList --custom /data/GRCh37_inclusion_list_v1.0.1.vcf.gz,IncludeVariant,vcf,exact,0,IncludeList --plugin SpliceAI,snv=/data/spliceai_scores.masked.snv.hg19.vcf.gz,indel=/data/spliceai_scores.masked.indel.hg19.vcf.gz --plugin REVEL,/data/revel_b37.tsv.gz --plugin CADD,/data/cadd_whole_genome_SNVs_GRCh37.tsv.gz,/data/gnomad.genomes.r2.1.1.indel.tsv.gz,/data/InDels_GRCh37.tsv.gz --fields Allele,SYMBOL,HGNC_ID,VARIANT_CLASS,Consequence,IMPACT,EXON,INTRON,Feature,HGVSc,HGVSp,HGVS_OFFSET,Existing_variation,STRAND,ClinVar,ClinVar_CLNSIG,ClinVar_CLNSIGCONF,ClinVar_CLNDN,gnomADg_AC,gnomADg_AN,gnomADg_AF,gnomADg_nhomalt,gnomADg_popmax,gnomADg_AC_popmax,gnomADg_AN_popmax,gnomADg_AF_popmax,gnomADg_nhomalt_popmax,gnomADe_AC,gnomADe_AN,gnomADe_AF,gnomADe_nhomalt,gnomADe_popmax,gnomADe_AC_popmax,gnomADe_AN_popmax,gnomADe_AF_popmax,gnomADe_nhomalt_popmax,gnomADe_non_cancer_AC,gnomADe_non_cancer_AN,gnomADe_non_cancer_AF,gnomADe_non_cancer_nhomalt,gnomADe_non_cancer_AC_popmax,gnomADe_non_cancer_AN_popmax,gnomADe_non_cancer_AF_popmax,gnomADe_non_cancer_nhomalt_popmax,gnomADe_non_cancer_popmax,TWE_AF,TWE_AC_Hom,TWE_AC_Het,TWE_AN,HGMD,HGMD_PHEN,HGMD_CLASS,HGMD_RANKSCORE,ExcludeVariant_ExcludeList,IncludeVariant_IncludeList,SpliceAI_pred_DS_AG,SpliceAI_pred_DS_AL,SpliceAI_pred_DS_DG,SpliceAI_pred_DS_DL,SpliceAI_pred_DP_AG,SpliceAI_pred_DP_AL,SpliceAI_pred_DP_DG,SpliceAI_pred_DP_DL,REVEL,CADD_PHRED --buffer_size 500 --fork 16 --no_stats --compress_output bgzip --shift_3prime 1'

[ainefairbrother]

Hi @kjwinfield,

Thanks for sending this through.

I think what's happening here is similar to issue #1867 and is a known bug. The --custom annotation columns are blank because --shift_3prime 1 shifts the allele that VEP uses for consequence calculation, but then VEP tries to look up that shifted allele in your custom annotation files, but these records are unshifted. As such, it won't be found and so VEP skips the custom annotation, leaving it blank.

Running without --shift_3prime 1 allows this variant to be annotated.

I hope this helps. Please let me know if you have any further questions.

Best,

@ainefairbrother

[kjwinfield]

hi @ainefairbrother,
Thanks for investigating. Is there a way that I can represent this variant in the custom annotation files to make it get annotated?
I have tried the representations below, and none of them work:

1	55505552	rs113330492	A	ACTG	.	.	ExcludeList=EXCLUDE
1	55505573	.	G	TGCG	.	.	ExcludeList=EXCLUDE
1	55505575	.	T	TGCT	.	.	ExcludeList=EXCLUDE

Turning --shift_3prime 1 off is not currently an option for us, as we need the HGVS c. notation to be right aligned for downstream interpretation
I've run it again with the above representations of the variant using v114.1, which the issue linked suggests is the fix (#1867 (comment)). The variant still was not annotated. Is there a timeframe for this bug to be fixed?

Best wishes,
@kjwinfield

[ainefairbrother]

Hi @kjwinfield,

Apologies for the delay in getting back to you.

Turning --shift_3prime 1 off is not currently an option for us, as we need the HGVS c. notation to be right aligned for downstream interpretation ...

--shift_3prime is only about where VEP positions the variant for consequence calculation, and does not affect the HGVSc notation.

However, if you still need to retain --shift_3prime 1 flag for other reasons, a work-around I could suggest is pre-processing your custom VCF files so that they are also shifted. This will mean that when VEP looks up the shifted variant in your custom files, it will find a match.

The fix for this is potentially quite complex, and as such, we currently do not have a defined timeline.

Please let me know if you have any further questions.

Kind regards,

@ainefairbrother