Why do half of my variants get rejected in picard LiftoverVcf?
You may not be providing support for picard, but at least the LiftoverVCF tool is referred to several places in this forum. I am working on two batches of bam-level data from the same project that have unhappily been aligned to b37 and hg19 respectively. My plan has been to lift over the hg19 files to b37 in the g.vcf stage and use GenotypeGVCFs on the full dataset. This is crucial to obtain a VQSR-worthy dataset downstream, which will then be 50-60 exomes. However, when I use picard LiftoverVCF with hg91tob37.chain, about 50% of variants are rejected due to mismatching reference alleles:
INFO 2016-03-14 13:58:26 LiftoverVcf Processed 11354836 variants.
INFO 2016-03-14 13:58:26 LiftoverVcf 0 variants failed to liftover.
INFO 2016-03-14 13:58:26 LiftoverVcf 5319041 variants lifted over but had mismatching reference alleles after lift over.
INFO 2016-03-14 13:58:26 LiftoverVcf 46.8438% of variants were not successfully lifted over and written to the output.
Any comments as to what might be the problem would be deeply appreciated!
It should be possible to leftover and GenotypeGVCFs together, right?