MARKER & MUTATION COMPARISON

Based on Doug MacDonald's 28 Feb 2004 Analysis, with several updates.  (See his full text below)

This article compares the markers used in the Family Tree DNA 12-25-37 marker test with those used in the Relative Genetics 26 marker test and provides estimated mutation rates for most markers. 

Background:  Our original Barton DNA Project testing was done at Relative Genetics (RG) and we are currently getting 22 of our 119 men tested on Panels 2 & 3 of the Family Tree DNA (FTDNA) 37 Marker Test.  Because of this, I am quite interested in the comparison of markers and mutation rates from one Testing Company to another.   This chart is compiled from FTDNA & RG public information, with heavy reliance on Doug McDonald’s deduction of Sorenson Molecular Genealogy Foundation data to obtain estimated marker mutation rates.  The color coding is explained in the notes at the bottom.

This is the “First Panel” of Family Tree DNA’s Test – Also Called their “12 Marker Test”

Locus

1

2

3

4

5

6

7

8

9

10

11

12

 

Average

DYS#

393

390

19/394

391

385a

385b

426

388

439

389-1

392

389-2

 

 

Rate

0.0012

0.0045

0.0016

0.0036

0.0033

0.0033

0.0005

0.0005

0.0045

0.0021

0.0016

0.0028

 

0.002458

This is the “Second Panel” of FTDNA’s Test – The 2nd part of their “25” Marker Test

Locus

13

14

15

16

17

18

19

20

21

22

23

24

25

Average

DYS#

458

459a

459b

455

454

447

437

448

449

464a

464b

464c

464d

 

Rate

0.0066

0.0014

0.0014

0.0005

0.0005

0.0045

0.0020

0.0028

0.0075

0.0035

0.0035

0.0035

0.0035

0.003169

This is the “Third Panel” of FTDNA’s Test – The 3rd part of their “37” Marker Test

Locus

26

27

28

29

30

31

32

33

34

35

36

37

 

Average

DYS#

460

H 4

YCAIIa

YCAIIb

456

607

576

570

CDY a

CDY b

442

438

   
Rate

0.0028

0.0036

0.0014

0.0014

??

??

??

??

??

??

??

0.0012

 

Unknown

Comments:

1.      FTDNA’s 25 marker test has a mutation rate of .002828, about 24% higher than RG’s 26 marker rate of .002285

2.      Unfortunately, SMGF did not include a number of FTDNA’s Panel 3 markers in their mutation analysis, so no speculation about the average mutation rate of Panel 3 can be made.

3.      FTDNA has 13 “fast moving” markers based on Doug’s mutation rate deductions and my presumption that “fast moving” means faster than .0028. 

4.      Note that FTDNA considers four markers that have no deduced mutation rates in Panel 3 as “fast moving” and also classifies 459a and 459b as “fast moving”, while Doug deduced the 459a and 459b mutation rates to be slow moving (.0014 each) from Sorenson’s data..

5.      FTDNA has 12 “slow moving” (less than .0020) markers: 5 in Panel 1, 4 in Panel 2 and 3 in Panel 3

 

The following two grids are the Relative Genetics “26 Marker Test”

DYS#

385

385II

388

389

389II

390

391

392

393

19/394

426

437

438

Rate

0.0033

0.0033

0.0006

0.0021

0.0028

0.0045

0.0036

0.0016

0.0012

0.0016

0.0005

0.0020

0.0012

 

DYS#

439

447

454

455

460

461

462

1BO7

YCAIIa

YCAIIb

A 10

C 4

H 4

Average

Rate

0.0045

0.0045

0.0005

0.0005

0.0028

0.0028

0.0005

0.0013

0.0014

0.0014

0.0045

0.0028

0.0036

0.002285

Comments:

1.      RG has 8 “fast moving” markers with a mutation rate of more than .0028.

2.      RG has 12 “slow moving” markers with a mutation rate of less than .0020 (shown in italics)

Notes:

1.      Markers shown in red are “Fast Mutating” according to FTDNA. 

2.      Markers shown in blue are other fast moving markers (faster than 0.0028, which is my classification decision)

3.      Mutations are from the analysis made by Doug McDonald of Sorenson Molecular Genealogy Foundation’s database, posted at Gen-DNA-L on Feb 28, 2004, with an update to me by email on April 7, 2004.  Note that Doug listed quite a number of cautions and caveats in his email.  A complete copy of his posting is provided below

4.      The cells shown in green are included in the FTDNA test, but not in the RG 26 marker test

5.      The cells shown in yellow are included in the RG 26 marker test, but not in the FTDNA tests.

6.      If there are any errors, they are mine.  Please advise and I will correct them.  Terry Barton - April 8, 2004

Here is Doug’s first email posting at

http://archiver.rootsweb.com/th/read/GENEALOGY-DNA/2004-02/1078025386

GENEALOGY-DNA-L Archives

From: doug mcdonald
Subject: [DNA] Separate mutation rates of each marker
Date: Sat, 28 Feb 2004
19:29:46 -0800 (PST)

I have deduced the mutation rates used by Sorensen.

 

CAVEAT: this is a very tricky procedure which is very sensitive to a certain normalization constant. It also depends on them using the formulas 20a and 20b of Walsh's paper. Because of the great sensitivity to parameter errors, the numbers that are farthest from the average are subject to the worst calculational errors, especially the SMALL ones.

 

In other words, BEWARE of these numbers. However, because of the way it works, they should come out in the correct order, smallest to largest.

 

ALSO NOTE: we have NO IDEA how Sorensen obtained the numbers. They do say they are using the multiple alleles model. We have no idea what THEIR error bars are. We have NO IDEA how accurate my back calculation is.

 

In other words, BEWARE of these numbers. However, because of the way it works, they should come out in the correct order, smallest to largest.

(Duplicate paragraph is not typo!)

 

 

NOTE: because of the way they are handling the multiple copy markers (385, 389, 459, YcaII) there is no info on the separate pieces.

 

But here they are:

 

385    .0066

388    .0006

389    .0020

390    .0045

391    .0035

392    .0015

393    .0012

19     .0015

426    .0006

437    .0020

438    .0011

439    .0045

447    .0045

448    .0028

449    .0075

454    .0006

455    .0006

458    .0066

459    .0028

460    .0028

461    .0028

462    .0005

ggaat1b07      .0013

ycaii          .0028

ygataa10       .0045

ygatac4        .0028

ygatah4        .0036

 

the overall average is .0028 ... averaging over the two marker ones only once.

 

The overall average is .0029 if you average over each of the double marker ones twice.

 

the average over the FTDNA 12 marker set (averaging over the two-marker ones twice) is .0029

 

the average over the 21 ones common to this and the FTDNA25, using multiple copy ones twice, is .0031 (note that this leaves out 464)

 

Doug McDonald

And here is Doug’s April 7, 2004 email to me with clarifications and updates:

-----Original Message-----

From: doug mcdonald

Sent: Wednesday, April 07, 2004 4:31 PM

To: Terry Barton

Subject: RE: Marker Mutation Analysis

 

--- Terry Barton

Doug, thanks!

Let me make sure that I am understanding you correctly.  Do I also split the

mutation rate for 389 and 389ii, getting a pair of .0010s

No  .... 389-I is .0021 and 389-II is .0028

and for YCAIIa and b, getting a pair of .0014s?

yes

I am also reading at the RG site  that they are calling 464 the marker and the a,b,c,d as alleles.  I understand you to mean that the mutation rate for R1b at 464 is about .016 or .004 for each of the 4 reported alleles.  Is that correct?

yes .... if you list them separately, like FTDNA does, the number for each is the realm of .004

This is extremely rough, and I'm not sure I believe it ... it could be .003. It seems to vary a lot for different datasets. Maybe you should use .0035.

I'm attaching a list of my latest Sorenson numbers ... these are per marker, just like FTDNA lists them.

Doug

****************

Here is a second look at the Sorenson rates. None have changed from my first list by a significant amount. The slowest ones dropped a bit, and in fact they may actually be as low as .0004. They are hard to back-calculate.

Again, this is Sorenson’s data, not mine. I merely reverse engineer assuming Walsh’s infinite alleles model. I have done several different methods of

checking and all agree.  The error bars are of course known only to Sorenson.

Hopefully they will publish soon.

Marker Rates                                                                                                                                                        

385    .0033 (*)

388    .0005

389I   .0021

389II  .0028

390    .0045

391    .0036

392    .0016

393    .0012

19     .0016

426    .0005

437    .0020

438    .0012

439    .0045

447    .0045

448    .0028

449    .0075

454    .0005

455    .0005

458    .0066

459    .0014 (*)

460    .0028

461    .0028

462    .0005

ggaat1b07      .0013

ycaii          .0014  (*)

ygataa10       .0045

ygatac4        .0028

ygatah4        .0036 

Note: the values marked by a (*) have been divided by two from what Sorenson uses.

This is because they are double markers. What Sorenson is doing is counting the pair as one marker. If either changes they count this as one mutation. If you do

this, you need to double the rate I show here. If you count the marker twice, as two markers, the best you can do is use the rates I show here, which are half

what Sorenson uses. They could, of course, actually be two different rates which average the value I give in this list. The two 389 values are listed separately

and does not suffer this ambiguity.

Doug

And here is Doug’s May 31, 2004 email with more clarifications:

-----Original Message-----

From: Doug McDonald [mailto:somerledsdna@yahoo.com]

Sent: Monday, May 31, 2004 12:52 PM

To: GENEALOGY-DNA-L@rootsweb.com

Subject: Re: [DNA] Marker Mutation Rates and other Tools

 

--- Terry Barton <terry@bartonsite.org> wrote:

I noticed the discussion tonight about YCAII a and b tonight which touched on their mutation rate. According to the work that Doug McDonald teased out of Sorenson's website, the mutation rate for marker YCAII is about .0028, which means that the rate for YCAIIa and YCAIIb is about .0014 each,

Please note that the decision to simply give the two values equal rate, i.e. deciding to divide by 2) was an ARBITRARY decision, because I really wanted to do SOMETHING. Since these markers trpically have similar values, it's probably not terrible.

For markers like 464, where values differ a lot, it is a bad idea, and I actually have data to show that the higher numbers mutate faster, especially >16. Somerled's 12-15-15-16 is extremely stable.

Doug McDonald