selection of reference genome

Question

[Tsin-Lau]

hello everyone, I got a vcf file with variation called using hg38 as reference genome. I wonder what would happen if I use hg19 as reference genome to annotate these variants. Would it be OK or get wrong? Thanks!

Answer 1

[Brian Bushnell]

Alignment, variation-calling, and annotation should always be performed on the exact same reference. Not a similar one. If you want to use a different reference, start anew from alignment. It's also possible to use LiftOver but you'll get a different answer.

[Tsin-Lau]

Thank you for your answer! Another small question: by saying 'a different answer', do you mean the LiftOver result is not that accurate?

[Brian Bushnell]

A lifted-over VCF will typically be 'mostly' the same as one generated from the target genome, especially in well-established regions of the reference that don't change much from version to version. But it's mainly useful when everything between two builds is identical except for the coordinates; when there are structural differences, or variants that span regions where a different major allele was chosen, there will be differences, and a better approach is to redo everything starting from alignment.

[Tsin-Lau]

I see, thanks.