Archiver > GENEALOGY-DNA > 2007-03 > 1174753590

From: Vincent Vizachero <>
Subject: [DNA] Misclassification as U4
Date: Sat, 24 Mar 2007 11:26:30 -0500

I've been working on mtDNA haplogroup U3 over the past couple weeks,
and have concluded that a particular group belonging to U3a are being
systematically misclassified as haplogroup U4.

According to a heuristic proposed by Richards et al. (2000), and used
by FTDNA it seems, anyone in haplogroup U with 16356C is assigned to
haplogroup U4.

However, there appears to be a subclade of U3a that also has 16356C
in addition to the "typical" U3a polymorphisms of 16343G and 16390A.
For this subclade (about 20-25% of all U3s), the Richards/FTDNA
heuristic fails to assign the correct haplogroup.

If you or a member of your project is classified in mtDNA haplogroup
U4 but has the HVR1 polymorphisms of 16343G and 16390A, then you are
almost certain to belong to haplogroup U3 instead. This can be
further confirmed by the presence in HVR2 of 150T (which U4 almost
never has) and the absence of 195C (which U4 almost always has). Any
"U4" with 16343G and 16390A is welcome to join the U3 project.

I base my conclusion on the following evidence.

1. Maca-Meyer published a confirmed U3 haplotype (GenBank accession
number AF381982) that has the 16356C polymorphism, but there is no
confirmed/published U4 haplotype with 16343G, 16390A, or 00150T.

2. 16356C is more prone to parallel mutations than 16343G, according
to studies by Allard et al. (2002) and Bandelt et al. (2006), and is
thus a less reliable predictor of haplogroup.

3. I performed an analysis of all U3 and U4 entries which included
HVR2 results that I could identify in the mitosearch database. A NJ
tree (confirmed by parsimony ratchet bootstrapping) clearly shows the
phylogeny of U3 and U4. This link to a PDF version of the tree shows
U3 clades in blue (with the misclassified group in red) and actual U4
clades in black. CRS and U Root were used as outgroups.

4. In addition, I published a MJ network that includes U3 haplotypes
from mitosearch and SMGF that shows another view. U3a is yellow (with
the misclassified U4s in red) and U3b is in blue. This network is

No HVR1-only schema will be 100% accurate in indentifying
haplogroups but looking at 16343 and 16390 in addition to 16356 will
produce much more accurate assignments than only looking at 16356.
In fairness to FTDNA, no one has taken a serious look specifically
at U3 or U4 so their reliance on Richards' scema is not surprising.
And since it was Richards, not FTDNA, who has made the ultimate
mistake it is up to FTDNA's advisors to evaluate the evidence (which
I have presented them).


This thread: