bcftools norm --check-ref not flagging alleles

[dwaggott]

I have a vcf generated from snp chip data via plink. The vcf contains alleles I and D instead of the actual sequence. I tried to use --check-ref xw to remove them but no filtering/warning seems to happen. Is this expected?

bcftools norm -d any -f GRCh37.fa -N -c wx -O z tmp.vcf.gz > tmp2.vcf.gz

[dwaggott]

A quick workaround is to use standard filters for problematic REF/ALT alleles. This won't catch the SNPs with incorrect REF/ALT due to chip design mapping issues though. That -c flag would be helpful.

bcftools view -e 'REF="N" | REF="I" | REF="D"' tmp.vcf.gz | bcftools norm -d any -f GRCh37.fa -N -O z > tmp2.vcf.gz

[pd3]

This should solve your problem pd3@63566bf, let me know in case of problems.

You may also want to check the fixref plugin for fixing REF/ALT mismatches.