maf2vcf.pl doesn't recognise N bases

[shadihames]

Hi, I'm working with MAF files that were generated by another person from Caveman/Pindel calls and I'm trying to convert them to VCF files to work with another tool. For some of the samples I get the error:

ERROR: MAF line 2040 (at 22:19178130) contains invalid alleles in Tumor_Seq_Allele or Match_Norm_Seq_Allele columns!

I've found this in a few files and it all seems to go back to the same position that has this Ref sequence:

AAAGATCACTGNTCACAGATCACCATACCATNTNNNGNNCN

I'm presuming that the issue is due to the N bases in the sequence, is this something that can be worked around, or would I have to pull our positions with N bases prior to converting with maf2vcf.pl?

Command:

perl maf2vcf.pl --input-maf sample.txt --output-dir sample --ref-fasta 37_decoy.fasta

Thanks!

[drychkov]

I've got the same error while using maf2vcf to convert maf to vcf.

@shadihames, did you figure out any workaround?

[sssimonyang]

I am trying to convert mc3.v0.2.8.PUBLIC.maf to vcf file and also came across this problem. And I just ignored those variants using awk.

sed 's/\r//' mc3.v0.2.8.PUBLIC.maf |awk 'NR==1 || ($13 ~ /^[AGCT\-]*$/ && $18 ~ /^[AGCT\-]*$/) {print $0}' > mc3.v0.2.8.PUBLIC.fixed.maf

perl ~/software/vcf2maf-1.6.21/maf2vcf.pl --input-maf mc3.v0.2.8.PUBLIC.fixed.maf --output-dir vcf --output-vcf mc3.v0.2.8.PUBLIC.vcf --ref-fasta ${hg19}

Hope to help you