PDA

View Full Version : Loss of confidence in mtDNA testing for genealogy: multi-generational mutations



Timothy
03-04-2015, 07:57 AM
The two separate multi-generational mtDNA lineage tests that I've done have come back with so many mutation differences from one generation to the next that it's caused me to conclude that mtDNA tests should be considered virtually worthless for genealogically relevant matching to the database.

In one line, haplogroup D4a1f (Japanese), I tested a mother, her son, and her daughter's son. They all have so many Coding Region differences that none show as a FMS match to the other! In addition, the grandson has an extra mutation in HVR1, so he isn't a match to his grandmother and uncle in HVR1 and HVR1/HVR2 comparisons either!

In the other line, H3b1b1 (Irish), I tested a mother, daughter, and granddaughter. The daughter has 1 Coding Region difference with her mother, and the granddaughter has 2 HVR2 differences with her mother and grandmother, so she's 3 differences to her grandmother!

Am I the only one to have such mind blowing results? Until this, I'd always assumed that mtDNA results remained fairly constant for many generations!

lgmayka
03-04-2015, 11:02 AM
Did you order mtFullSequence from FTDNA in all cases?

rod
03-04-2015, 06:42 PM
What were the dates of the tests? I wonder if the "next generation" tests are less reliable than the old tests. Prior to April 2013, Family Tree DNA performed mitochondrial DNA (mtDNA) tests by direct (Sanger) sequencing then they switched to Illumina MiSeq.
My vintage 2008-2010 Sanger FMS tests produced exact parent-child matches. Illumina tests of the same people at 23andme gave random different results on some mtDNA SNPs including one that affected Haplogroup assignment.

Little bit
03-04-2015, 07:29 PM
If the full mitochondrial sequence wasn't so expensive, I'd test one of my kids or mom. Very interesting, and disappointing! I ran my group's 23andme raw data through Pike's 'Search for differently reported snps' tool and here's the results:

4 J1c3i's - me, my mom, my son, my daughter - no differences found - 2 are v3's, 1 is v2, and I am v2/v3 upgrade

2 T2b7a1's - husband and his mom - 1 difference but it's probably not legitimate.
16257 i3001755 C or T Mom is T, son is A. My husband is a v2, his mom v3, and 23andme probably changed orientation or snp id over the years

2 K1a4a1d's - my grandpa (v2/v3 upgrade) and his brother (v3) - 1 difference and this looks like a real discrepancy, not sure if it's a bad call by 23andme or a bona-fide mutation?
10619 i4000831 C or T My grandpa is T, his brother is C. 10619T shows up as an extra on my grandpa's James Lick analysis.

Unfortunately I don't have anyone besides me mtdna tested at FTDNA but if they ever lower the price of the FMS, I may consider testing some parent/child pairs to see if they are matches to each other. Thanks for bringing this to our attention!

Huntergatherer1066
03-04-2015, 11:54 PM
My advisor from graduate school was very interested in mtDNA mutation rates; some of her findings were suggesting that mtDNA may mutate more rapidly than thought. Here is one paper she published, I got to help present a follow-up poster on it at an AAPA conference. Unfortunately she only tested HVR1; it would be interesting to revisit the study doing full mitochondrial sequencing:

http://anthropology.usf.edu/faculty/personal/publications/Madrigal%20mutation.pdf

GailT
03-05-2015, 02:35 AM
I've seen a few cases with 3 or differences in 1 or 2 generations, but this is usually an artifact of the way FTDNA counts differences and heteroplasmies. It is much more common for there to be no mutations in thousands of years. If you are willing to share the complete results I can check to see how they counted the differences.

Timothy
03-05-2015, 06:22 AM
Did you order mtFullSequence from FTDNA in all cases?

Yes, they are all mtFullSequence from FTDNA. I also did a mtPlus on the grandson with the HVR1 difference to his grandmother and uncle, to verify the result, 4 yrs later, and same result.

Timothy
03-05-2015, 06:31 AM
What were the dates of the tests? I wonder if the "next generation" tests are less reliable than the old tests. Prior to April 2013, Family Tree DNA performed mitochondrial DNA (mtDNA) tests by direct (Sanger) sequencing then they switched to Illumina MiSeq.
My vintage 2008-2010 Sanger FMS tests produced exact parent-child matches. Illumina tests of the same people at 23andme gave random different results on some mtDNA SNPs including one that affected Haplogroup assignment.

Good question, Rod!

The Japanese grandson and the Irish grandmother and daughter were 3 or 4 yrs ago. The Japanese grandmother and her son, and Irish granddaughter, were just completed in the last couple months.

The Japanese grandson was a proxy for my wife, who refuses to test on General Principles (that I spend too much money doing this!). He has 1 Chinese HVR1 match. Imagine my shock after testing grandmother and uncle, as they have a slew of Chinese, Korean, and Japanese matches!

Timothy
03-05-2015, 06:44 AM
My advisor from graduate school was very interested in mtDNA mutation rates; some of her findings were suggesting that mtDNA may mutate more rapidly than thought. Here is one paper she published, I got to help present a follow-up poster on it at an AAPA conference. Unfortunately she only tested HVR1; it would be interesting to revisit the study doing full mitochondrial sequencing:

http://anthropology.usf.edu/faculty/personal/publications/Madrigal%20mutation.pdf

Very interesting indeed! And yes, a FMS version could be most enlightening!

Timothy
03-05-2015, 06:48 AM
I've seen a few cases with 3 or differences in 1 or 2 generations, but this is usually an artifact of the way FTDNA counts differences and heteroplasmies. It is much more common for there to be no mutations in thousands of years. If you are willing to share the complete results I can check to see how they counted the differences.

Thank you for the offer! I won't be able to do this tonight, but will compile the info ASAP! And, yes, some of the differences are heteroplasmies (in both lines).

GailT
03-06-2015, 04:09 AM
Thank you for the offer! I won't be able to do this tonight, but will compile the info ASAP! And, yes, some of the differences are heteroplasmies (in both lines).

I some cases FTDNA counts a common heteroplasmy such as 16093Y as 2 steps, but it should be ignored, and I would count it as 0 steps. Multiple insertions or deletions are also counted as multiple steps, but they should also be counted as 0 steps, or in some cases as 1 step.