No announcement yet.

mtDNA Discrepancies: FTDNA versus 23andMe

  • Filter
  • Time
  • Show
Clear All
new posts

  • mtDNA Discrepancies: FTDNA versus 23andMe

    I recently received my mtDNA results from a mtFullSequence test I had done here at FTDNA. I’ve compared those results to my 23andMe Complete Edition results. I’m surprised by the number of discrepancies. To my eyes, most of the discrepancies, if not all, seem to arise from 23andMe errors but I’m not sure. BTW, I know 23andMe uses Yoruba and FTNDA uses CRS and I’ve taken this into account.

    As far as I can tell, FTDNA does not provide the ability for customers to download the entire mtDNA results file. It seems as though the results are only provided in a form that identifies mutations, insertions and deletions that don’t match the CRS genome. Because of this, I can’t tell how many, if any, “No Calls” are in my FTDNA results. I’m assuming that if FTDNA doesn’t report a position, they called that position as identical to the CRS genome? Is that correct or does FTDNA have “No Calls” and just ignore them?

    As for comparions of FTDNA results with 23andME results, I’d appreciate it if any of you can offer a suggestion as to why so many 23andMe results either don’t match or make sense. If I’m wrong about this, I’d like to know. To me it looks like FTDNA is correct and 23andME is wrong but I’m not certain of that.

    In my 23andMe mtDNA Raw Results, I’ve identified 10 mismatches with FTNDA results, 3 duplicate SNPs and 15 "No Call" locations. They are summarized at this URL:

    Of course, the big issue is the 10 'errors'.

    If anyone would like to see my compete set of 23andMe RAW mtDNA results in annotated format, you can look here:

    I have also posted my FTDNA results here:

    As an aside, my mtDNA Haplogroup appears to be “H*”. I notice some mtDNA “H” haplogroup forums restrict membership to H’s who are 263G and 315.1G. What is the special significance of those two positions?

    Hoping for some feedback.

    Thanks, Mardon
    Last edited by Guest; 11 March 2010, 12:44 PM. Reason: Fix Typos

  • #2
    I haven't checked my mtdna in 23andme, but 23andme's technology doesn't work well for mtdna and gives a lot of no calls. Part of the reason, I believe, is that it uses as markers the other positions near the position being analyzed. If there is a mutation or something similar (eg heteroplasmy) nearby, the reading can get confused.

    So I would go with FTDNA.

    As a curio, I did both ftdna and another company, Argusbio, that is not offering the test anymore. There one difference between the two, FTDNA had one additional mutation.

    As for 263, it is a mutation specific of H2 and CRS in particular.



    • #3
      FTDNA used to provide the FASTA file, but they've removed the links in the results pages pending an upgrade to website security. Hopefully, access to those files will be restored to us soon.


      • #4
        No-calls are not an issue for FTDNA. They are actually sequencing the entire mtDNA molecule, while 23andMe is probing for a couple of thousand SNPs. This can be tricky, since any variations in the vicinity of a SNP can affect the call. For instance, 23andMe probes for 263, but a person with a very unusual mutation at 262 (not included in the 23andMe set) got a no-call for 263. The mutation at 262 prevent the probe from binding to the 263 site.

        Last October 23andMe began calling more sites, some of which have turned out to be problematic. I suspect those locations will no longer be reported in the next update. I'd suggest just setting your analysis aside for the time being.


        • #5
          Hi Ann,

          Thanks for your reply. You’ve raised some interesting points that I hadn’t considered and still don’t understand.

          I did know that FTDNA sequences the entire mitochondrion. I did not know that “No Calls” are impossible when doing a full sequence test. I guess I don’t understand enough about how the actual sequencing is performed to understand this. I don`t understand how the allele at one location can affect the call at a nearby location.

          I have read on the 23andMe website that they use an Illumina HumanHap 550+ BeadChip to make their SNP calls. I also knew that they don’t sequence the entire mitochondrion. Their website states, "... the HumanHap550+ Genotyping BeadChip from Illumina analyzes more than 2000 SNPs across the entire mitochondrial genome." According to that, they sample less than 15% of the mtDNA positions.

          I found another source that states, "Illumina reported that the rate of Mendelian inconsistencies of Illumina Human-Hap550 beadchip is 0.06%."

          If I`m understanding this information properly, 0.06% times 2,000 positions indicates an average of 1.2 errors per 2,000 calls.

          Does FTDNA use a similar type of BeadChip to peform the sequencing? If so, doesn‛t 0.06% times 16,600 positions indicate an average of 10 errors in a full sequence of the mtDNA? Is this the sort of error rate that FTDNA would expect in their full sequence mtDNA results or do they sequence any non-CRS locations twice to be certain they are not errors?

          Any more details you can provide to help me understand this better would be appreciated.

          For now, I`m assuming the FTDNA results are 100% correct and the mis-match with 23andMe is all the result of 23andMe errors.

          Thanks again, Mardon
          Last edited by Guest; 12 March 2010, 05:32 PM.


          • #6
            No, sequencing at FTDNA does not use a BeadChip. It "reads" the bases one at a time, and it doesn't care what base comes next. The BeadChip is probing for particular patterns that match a whole series of bases. For example,


            is known to vary sometimes in the 4th position


            The BeadChip contains both versions (the variant plus some surrounding bases), and your mtDNA will attach to one or the other if it matches the whole string. If your mtDNA also has a mutation at the 8th position, that spoils the match.

            Thus the BeadChip can only look for specific patterns that have already been identified, but sequencing can find mutations that have never been reported before.


            • #7

              The information you’re providing is extremely interesting. I did not know about the different approaches used by the BeadChip and the FTDNA method of directly reading the individual bases. Is there a name for the chip/technology that FTDNA uses to “read” the individual bases?

              Now that FTDNA is introducing “Family Finder”, what method will FTDNA use to read the SNPs on the autosomal chromosomes? Do the same issues apply that you’ve described for the mtDNA? In other words, should FTDNA’s calls used for “Family Finder” be more accurate than the calls made by 23andMe and used for their “Relative Finder”?

              Many thanks for your help,



              • #8
                Family Finder will use the Affymetrix chip.

                In a move that puts it in more direct competition with personal genomics companies such as 23andMe and deCODEme, the genetic genealogy testing company Family Tree DNA announced today that it will offer a large-scale autosomal test for genealogical purposes.… Continue reading →


                • #9
                  Thanks Rucksack,

                  Do the Affymetrix GeneChips use the same technique that Ann described for the Illumina BeadChips, that is, of looking at patterns and then getting confused by previously unreported nearby mutations?

                  Are the Affymetrix GeneChips in anyway superior or inferior to the Illumina BeadChips?

                  Will FTNDNA now start using the Affymetrix GeneChip for testing mtDNA rather than the method that Ann said they have previously used of looking at individual bases?

                  Sorry for so many questions but I


                  • #10
                    I don't know about the differences for the Affy chip, and have no idea which is considered superior, or if that would even be a way to compare them.

                    No, the FGS will still be done the way it is currently done at FTDNA.

                    Another thing is that FTDNA will "scrub" the medical data from the results prior to providing the data to it's customers of the new Family Finder product.

                    The Y, X and mtDNA will not be a part of the Family Finder product either, I believe, whereas 23andMe does report them.


                    • #11
                      Originally posted by rucksack View Post
                      Another thing is that FTDNA will "scrub" the medical data from the results prior to providing the data to it's customers of the new Family Finder product.
                      Thanks again for your reply. How could FTDNA “scrub” the medical data from the results? The FTDNA Family Finder website contains this statement, “Your raw data file is freely available for download.” If that’s true, then the raw data can be run through a 3rd party analysis program, like Promethease, to get a medical interpretation of the results. I can understand that FTDNA won’t be reporting any medical results but I don’t see how they can prevent RAW results from being used as input for 3 rd party analysis. Am I missing something?


                      • #12
                        They will remove the data prior to making it available to you.


                        • #13
                          Here I am H7a, on 23andme I'm H7a1. It lists the same mutations for me though. 23 lists 16261 as being a marker for H7a1, and it's right there on my HVR1 here.


                          • #14
                            I just ordered 23 and me. when I get results i'll compare to my full sequence of mito dna I got from ftdna and post if there are any differences. at least its good insurance my sample wasn't swaoped for someone elses.

                            on 23 and me, is there a way to print an entire
                            entire report? i'd much rather read one hard copy from the couch than follow multiple links on the computer moniter. also, where do I find the instructions on how to download raw data from 23 and me in one big zip file. i've read that's offered but when I go to 23 and me faq all it says is for raw data to go to browse-and then click on eaxh individual chromosome, which is not what I want. thanks if anyone knows these answers and can post.

                            I hope someone also picks up interesting issues in this thread of family finder Vs 23 and me and more on whether med info is really deleted in raw data by removing relevent locations.


                            • #15
                              I had downloaded my FTDNA raw data to Promethease and there was plenty of medical data. I think FTDNA just does not report the medical information themselves but it is in our raw data. Two weeks ago my full brother gave and submitted a DNA sample to 23&ME and we are anxiuosly awaiting his results. I would like to compare his to mine. Also from Promethease I can compare some of his genes to mine medically. We figure we should have a lot of overlap and so his higher RF results would be mine too if I tested. Either way his relatives are mine genealogically on 23&Me and mine are his on FTDNA since we have the same parents. I know that he still would have some genes I don't and vice versa since only identical siblings are 100% related. Unluckily the cost is too high to have us both test with both companies.