You can read this thread – De novo calls on hemizygous X variants – for my slow realization that the answer to that question is “no” (since the caller looks for situations where the child’s genotype is het, and the VCF won’t report variants that way in hemizygous regions) and for the way I modified the caller to help it work.