New to using liftOver and working with vcf files generally:
I ran liftOver on data gathered from the PsychChip array to lift over from GRCh37 to GRCh38, and got only about 50% of variants lifted over. Most of the rejected ones had "MismatchedRefAllele" as their reason for rejection.
I was able to use PLINK previously to filter out indels for the Multiethnic array from Illumina, along with keeping the allele order, and I was able to get 96% of variants lifted over. I used the same commands to prepare the PsychChip vcf for liftOver, but had many more rejected.
Anyone have similar problems/know how to fix this? Thanks!
where does the VCF come from ? may be you could swap REF/ALT using
bcftools norm
.thank you! I will try this.
The VCF comes from PLINK.
Did you run Picard/LiftoverVcf with option
–RECOVER_SWAPPED_REF_ALT
? That might take care of the SNPs (but not the indels). Overall, as Pierre Lindenbaum suggested, the most likely culprit might be that your reference and alternate alleles are randomly swapped. How did you convert the PsychChip array data to VCF? Did you consider using BCFtools/gtc2vcf?Thank you! I did use –RECOVER_SWAPPED_REF_ALT. I used plink originally to convert from plink to vcf, which I know can swap alleles, so I used keep allele order, but it may not have worked. I'll try using your suggestions.