Why did two of my SNPs become one MNV?

[Paururo]

My variant caller reported two separate SNVs a couple of bases apart. Both are
real calls with good depth. In the get_MNV output they are on a single row,
typed as an MNV.

Why are they merged, and which amino acid should I trust: the two the caller
implies, or the one get_MNV reports?

[Paururo]

Because the amino acid depends on the codon as a whole, and not on either
change on its own.

Take a codon reading CGT, which codes arginine:

Read as	Codon	Amino acid
Reference	CGT	Arg
Position 1 alone, C>T	TGT	Cys
Position 2 alone, G>C	CCT	Pro
Both together	TCT	Ser

An annotator that takes each substitution separately reports a cysteine and a
proline. Neither of those happens. The protein carries a serine, and that is
what the merged row is telling you: the row is a codon, not a position.

Merging is per codon and nothing else. Two changes in neighbouring codons stay
two rows, however close together they are.

One thing the merged row does not claim on its own is that both changes sit on
the same molecule. Sharing a codon and sharing a molecule are different
statements. Give it a BAM and it counts the reads carrying both instead of
assuming, which is what the linkage columns are for:
https://pathogenomics-lab.github.io/get_MNV/linkage/