#Limitations

assign is not a haplogroup classifier. It reports the haplogroup at the smallest symmetric difference. It:

• ignores which positions were sequenced, so an unobserved position and an absent one count the same
• weights every difference alike, where a classifier weights by how phylogenetically informative a position is and discounts unstable ones
• applies none of the phylogenetic reasoning, alignment rules included, of EMPOP's or HaploGrep's classifiers

Treat the result as a starting point and check it with path and haplotype.

Range mismatch is unhandled by design. Phylogeny haplotypes span the whole genome; observations usually cover the control region only. assign compares designations regardless, so a haplogroup defined by coding-region variants cannot be told apart from its neighbours by a CR-only observation, and the distance does not show that this happened. Check lineage.mtRangesAgree.

No phylogenetic alignment. EMPOP shifts indels to the position the mtDNA phylogeny prefers. Neither this deck nor lineage.align does, so a complex indel may be designated differently from how mitoLEAF or EMPOP write it, and then it will not match.

The tree moves. mitoLEAF adds haplogroups, removes variants and respells back mutations between releases. An assignment means something only together with the release that produced it. The deck does not record the release; the phm file states it in a leading comment.

No tree is bundled. testdata/tree.xml is a seven-haplogroup synthetic fixture for the tests and the example, not the phylogeny.

Whole-tree operations take seconds. parse takes ~0.5 s and assign ~2 s on the published tree (figures): fine interactively, too slow inside a loop over thousands of samples without restructuring.

Not a validated forensic tool. Neither this deck nor @deckyard/forensicgenetics has been through the validation any jurisdiction requires of software used in casework.