GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2006-04 > 1146317774


From: "William Hurst" <>
Subject: Re: mtDNA mutation rate [was Re: [DNA] Age of J2 in India]
Date: Sat, 29 Apr 2006 09:36:14 -0400
References: <24c.a5028e5.318360c6@aol.com> <REME20060428234951@alum.mit.edu>


Hi Ann and John and all,

After looking at John's mtDNA mutation rate calculations, I decided to see
what I could determine by just looking at my haplogroup K's. I used the
members of the K project instead of MitoSearch. Although the sample is
smaller, this eliminates any duplications and any errors possibly caused by
the by-hand entries into MitoSearch.

The K project has 127 members, but I only looked at the 78 with HVR1+HVR2
data. I have demonstrated in my earlier K surveys that the basic K haplotype
has six mutations:

HVR1: 16224C, 16311C, 16519C
HVR2: 073G, 263G, 315.1C

Curiously, no K in the project or on MitoSearch (the last time I looked) has
just those six mutations. So if this is considered the ancestral haplotype,
it is NOT the modal haplotype. I am assuming that the original K (Katrine)
had those six mutations. And yes, I know that these are just differences
from the CRS and that, for example, for position 263 it's really the CRS
which had the mutation. I have seen two dates for Katrine, 15,000 and 16,000
years ago, so I used 15,500.

Next I counted the additional mutations for each project member with HVR1/2
results. Since the HVR2 512 insertions and the 522-523 deletions always (as
far as I know) appear in pairs, I've counted each pair as only one mutation.
Following is a chart of the additional mutations:

Zero: 0
One: 4
Two: 16
Three: 18
Four: 18
Five: 14
Six: 6
Seven: 0
Eight: 0
Nine: 2

So, 286 additional mutations divided by 78 entries = 3.67 as the average
additional mutations per entry - with a high variance.

The entries with only one additional mutation were in two pairs; one pair
added 146C, the other 497T. Both these are known to be very early mutations.
In Dr. Behar's study, 146C is a defining mutation for both subclades K1c and
K2, and 497T is the defining mutation for K1a. The two entries with nine
mutations are also a matching pair. The results look more or less like a
bell curve, as could be expected.

Next, dividing years by mutations gives the answer: one additional mutation
per entry every 4223 years. Since all these are rough calculations, I'll go
with 4000 years. Dividing the other way gives 0.00024 additional mutations
per year per entry, or a .024% chance any K will have a mutation in a year.

I could look around to see how this compares to other calculations of mtDNA
mutation rates, or I could have my coffee. Coffee wins! Maybe somebody knows
the answer to that.

Bill Hurst

This thread: