Have you ever thought about the number of reads truly needed to call a tetraploid genotype accurately?

If the answer is no, neither had I. And I had spent quite some thought on how to impute them (albeit mainly for the scenario of a subset of fully called genotypes)...