Paternity evaluation when there is tri-allelism

Paternity calculation when there is a tri-allele pattern at a locus

It's a difficult situation, because to calculate anything in science it's necessary to know the physical nature of the phenomenon, that is, to have a mathematical model of the reality, and for the most part we do not know what tri-allelism is because there are several very different possibilities.

The only simple point I can make is that the DNA·VIEW kinship and parentage programs will not accept 3 alleles at a locus for a paternity calculation, so don't enter all three.

There are several phenomena that manifest as tri-allelism which are very different from one another.

If the child inherited a chromosome with two copies of the locus, then the discussion at dna-view.com/3band.htm is relevant.
If trisomy, then then a varation of the idea dna-view.com/placental.htm, involving chooseing 2 out of 3 alleles, is worth thinking through.
The extra allele might come from a completely different chromosome. This different model necessitates a different answer.
For an example I recommend the interesting detective story of AB Lane's paper “The nature of tri-allelic TPOX genotypes in African populations”, Forensic Sci Int Genet. 2008 Mar;2(2):134-7.
If the tri-allelism is a new mutation (that is, not found in a parent), that's yet another problem.

So what's a practical person to do? It the locus seems consistent with paternity under all explanations then maybe you can omit the locus from calculation with the justification that the evidence from it, whatever it is, must be positive. Just a thought.