Why does vcf2maf filter out CNV / Fusion/ ProcControl / RNAExonVariant ?

[lakhujanivijay]

I have a vcf file which have few rows with following type of data and I could observe that all of this is being filtered out.

grep -v "^#" sample1_pateint1.vcf | grep -Po "SVTYPE=\w+"  | sort | uniq -c
      1 SVTYPE=CNV
     94 SVTYPE=Fusion
      2 SVTYPE=ProcControl
      4 SVTYPE=RNAExonVariant

My command is :

vcf2maf.pl --input-vcf sample1_pateint1.vcf --output-maf sample1_pateint1.maf --ref-fasta genome.fa --filter-vcf 0 --vep-path /home/bioinfo/variant-effect-predictor-87-0/bin/ --custom-enst isoform_overrides_uniprot.txt --tumor-id sample1

Can you tell us why is this happening?

[ckandoth]

Except for CNV, those are not valid values for SVTYPE as defined in VCF 4.2 specs. See this line of code in vcf2maf to see how SVs are handled. Briefly, SVs of type BND, TRA, DEL, DUP, or INV are split into two VCF lines for two breakpoints and they are fed into Ensembl VEP as point mutations. This is hacky, but it was the only way to generate the Gene1::Gene2 format for gene fusions that the cBioPortal importer liked to parse. This is old hacky code, so YMMV. It looks like the cBioPortal deprecated the Fusion format in favor of a more structured SV format.