Variant method to return actual states

[hyanwong]

@szhan and others are finding it pretty inconvenient to deal with mismatching between the underlying integers in the genotypes array returned by sample_data.variants() and ts.variants(). How about if we provided a method on the two variant classes to return the encoded variation as a numpy string array. It would be inefficient for large-scale stuff, but I think it might save many errors in smaller-scale testing, etc. Something like the following would probably work for SampleData instances, and an equivalent function could be created for tskit variants. Hopefully making it a function would make it clear to the user that a potentially inefficient calculation was going on under the hood.

@attr.s
class Variant:
    """
    A single variant. Mirrors the definition in tskit.
    """
    site = attr.ib()
    genotypes = attr.ib()
    alleles = attr.ib()

    def genotypes_as_strings(self):
        """
        Returns the variants at this site as an array of strings: Note, however, that it is
        much more efficient to work with the underlying integer representation as
        returned by the ``.genotypes`` property.
        """
        return np.array(alleles)[genotypes]

[jeromekelleher]

This is useful all right, we use something like this in a bunch of places.

[hyanwong]

Adding to 0.3.1 as this is a trivial but useful addition

[benjeffery]

+1 on this as I got very confused writing the sgkit ancestral allele tests.

[hyanwong]

Over in https://github.com/tskit-dev/tskit/pull/2617 @jeromekelleher suggested we call this method .states()