I found some inconsistencies between VEP annotated VCF gnomAD fields and gnomAD actual DB.

I am using VEP v98 on a Debian 9 machine. dbNSFP v4.0a.

As example, variant 12-116398626-A-C (here gnomAD page) is reported with a genome AF of 0.378 (37.8%).

If I isolate this variant only from VCF file:

chr12   116398626   .   A   C   584.19  PASS AC=2;AF=0.333;AN=6;BaseQRankSum=-1.007;ClippingRankSum=0;DP=42;ExcessHet=3.9794;FS=0;MLEAC=2;MLEAF=0.333;MQ=60;MQ0=0;MQRankSum=0;QD=17.7;ReadPosRankSum=-0.279;SOR=1.022 GT:AD:DP:GQ:PL    0/1:7,8:15:99:230,0,232 0/1:7,11:18:99:386,0,236    0/0:9,0:9:24:0,24,360

and pass it through VEP with both --af_gnomad and dbNSFP gnomAD AF flags:

/home/enrico/vep/ensembl-vep/vep -v \ -i "${sample_vcf_normalized}.ch12_116398626.vcf" --format vcf \ --cache --dir_cache /media/enrico/vep_disk/vep_files/ --merged --use_given_ref \ --fasta /media/enrico/vep_disk/vep_files/fasta/GRCh37/Homo_sapiens.GRCh37.75.dna.primary_assembly.fa.gz \ --dir_plugins /media/enrico/vep_disk/vep_files/Plugins/ \ --assembly GRCh37 \ --offline \ --minimal \ --vcf \ --hgvs \ --no_stats \ --symbol \ --allele_number \ --pick \ --no_intergenic \ --gencode_basic \ --sift b \ --polyphen b \ --plugin SingleLetterAA \ --plugin dbNSFP,/media/enrico/vep_disk/vep_files/plugin_databases/dbNSFP/dbNSFP4.0a.gz,LRT_score,LRT_pred,GERP++_RS,MutationTaster_pred,MutationTaster_score,MutationAssessor_pred,MutationAssessor_score,FATHMM_score,FATHMM_pred,1000Gp3_AC,1000Gp3_AF,1000Gp3_AFR_AC,1000Gp3_AFR_AF,1000Gp3_EUR_AC,1000Gp3_EUR_AF,1000Gp3_AMR_AC,1000Gp3_AMR_AF,1000Gp3_EAS_AC,1000Gp3_EAS_AF,1000Gp3_SAS_AC,1000Gp3_SAS_AF,UK10K_AF,ESP6500_AA_AF,ESP6500_EA_AF,gnomAD_exomes_POPMAX_AF,gnomAD_exomes_POPMAX_nhomalt,gnomAD_genomes_POPMAX_AF,gnomAD_genomes_POPMAX_nhomalt,GTEx_V7_gene,GTEx_V7_tissue,Geuvadis_eQTL_target_gene,Polyphen2_HDIV_score,Polyphen2_HDIV_pred,Polyphen2_HVAR_score,Polyphen2_HVAR_pred,ExAC_AC,ExAC_AF,ExAC_Adj_AF,ExAC_AFR_AC,ExAC_AFR_AF,ExAC_AMR_AC,ExAC_AMR_AF,ExAC_EAS_AC,ExAC_EAS_AF,ExAC_FIN_AC,ExAC_FIN_AF,ExAC_NFE_AC,ExAC_NFE_AF,ExAC_SAS_AC,ExAC_SAS_AF,REVEL_score,REVEL_rankscore,clinvar_id,clinvar_clnsig,clinvar_trait,clinvar_review,clinvar_hgvs,clinvar_var_source,clinvar_MedGen_id,clinvar_OMIM_id,clinvar_Orphanet_id,CADD_phred,ExAC_Adj_AC,gnomAD_exomes_AN,gnomAD_exomes_AC,gnomAD_genomes_AN,gnomAD_genomes_AC,gnomAD_exomes_controls_AC,gnomAD_exomes_controls_AN,gnomAD_exomes_AFR_AF,gnomAD_exomes_AMR_AF,gnomAD_exomes_ASJ_AF,gnomAD_exomes_EAS_AF,gnomAD_exomes_FIN_AF,gnomAD_exomes_NFE_AF,gnomAD_exomes_SAS_AF,gnomAD_exomes_controls_AF,gnomAD_genomes_controls_AC,gnomAD_genomes_controls_AN,gnomAD_genomes_AFR_AF,gnomAD_genomes_AMR_AF,gnomAD_genomes_ASJ_AF,gnomAD_genomes_EAS_AF,gnomAD_genomes_FIN_AF,gnomAD_genomes_NFE_AF,gnomAD_genomes_controls_AF,genename \ --custom /media/enrico/vep_disk/vep_files/manual_annotations/clinvar/clinvar_20191118.vcf.gz,ClinVar,vcf,exact,0,CLNSIG,CLNREVSTAT,CLNDN,CLNDISDB \ --force_overwrite \ --af_gnomad \ -o "${sample_vcf_normalized}.ch12_116398626.vep.vcf"

I got an empty fields for both, as the variant wouldn't be in gnomAD!

chr12   116398626   .   A   C   584.19  PASS    AC=2;AF=0.333;AN=6;BaseQRankSum=-1.007;ClippingRankSum=0;DP=42;ExcessHet=3.9794;FS=0;MLEAC=2;MLEAF=0.333;MQ=60;MQ0=0;MQRankSum=0;QD=17.7;ReadPosRankSum=-0.279;SOR=1.022;CSQ=C|3_prime_UTR_variant|MODIFIER|MED13L|ENSG00000123066|Transcript|ENST00000281928|protein_coding|31/31||ENST00000281928.3:c.*445T>G||7285|||||rs66898998|1||-1||1|HGNC|22962||Ensembl||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||   GT:AD:DP:GQ:PL  0/1:7,8:15:99:230,0,232 0/1:7,11:18:99:386,0,236    0/0:9,0:9:24:0,24,360

Does anybody knows why??

Thank you in advance for any help!