After generating the filtered VCF we usually do decompose variants and left normalization.
In contrast snpEff tool explains that in order to report variant annotations consistent with HGVS, variants need to be re-aligned according to each transcripts strand (most 3-prime coordinate of transcript) and then to calculate the annotation. I appreciate anyone helping me with my doubts below.
1) Is this re-aligning step happening by default? There's a option to deactivate this (no_shift_hgvs) so I assume it is the default action in tool.
2) Decomposing variants in a VCF before left-normalization, results in splitting VCF lines so each line has one variant. But in a case where a bi-allelic site has for eg. two alter alleles and the genotype is called as 1/2 (e.g ref is A and two alleles are G,T and the site has genotype G/T), decomposing will fail to retain this variant. Am I correct?
Thanking in advance