Announcement

Collapse
No announcement yet.

mtDNA Results from 23andMe

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • mtDNA Results from 23andMe

    Last year, when 23andMe reduced their price for Christmas, I paid for my dad to be tested through them. I knew they didn't test Y-STRs, but I already had my results from FTDNA. He's interested in genealogy (but not obsessed like me ), and I thought the health aspect would be useful too.

    We just got the results back yesterday, with mixed feelings. The Y-DNA haplogroup was N1c1*, so I am indeed his son. There are several RF matches, but their interface doesn't display the total and longest shared segment in cM. They only list a percentage and number of segments shared (but how long?).

    The most useful information for me was the mtDNA results. I was hoping to add him to mitosearch for possible matches. However, the format 23andMe uses isn't easily readable like FTDNA's. On their website all they show is the haplogroup. (Their descriptions are all pretty simplistic, but I already knew where to find better, more detailed descriptions.) You have to download the raw data for any further information. Then, instead of showing mutations vs. the CRS, it's merely a long list of positions and values.

    I had no desire to manually compare it to the CRS, so I checked out the various online tools, and James Lick's mthap did the trick. (If you haven't tried it before, you really should. FTDNA typed me as a K, but mthap was able to type me as essentially a perfect fit in K2b1a.) Besides checking your mtDNA against deeper sub-clades, it shows the exact steps, SNP by SNP, it uses to traverse the phylotree. And it provided the mutations vs. CRS, right at the top of the results page.

    Unfortunately, it also added a major caveat:
    IMPORTANT NOTE: The above marker list is almost certainly incomplete due to limitations of genotyping technology and is not comparable to mtDNA sequencing results. It should not be used with services or tools that expect sequencing results such as mitosearch.
    At first I didn't quite understand this, but figured it out after further reading here and on DNA-Forums. Mitosearch, and other mtDNA matching, depends on complete HVR1/HVR2 sequencing. 23andMe tests enough SNPs to type you, but it's nowhere near complete, even on those two regions. It's possible you have other mutations 23andMe doesn't test. So if you input those results on mitosearch, GD with any matches wouldn't necessarily be correct.

    While it was useful to test with 23andMe to compare their product with others, I don't think I'll use their services again. They weren't initially setup for genealogical DNA testing, and it shows. Many stories on this forum detail how the response rate when contacting matches there is pretty low. Some of their customers have no interest in genealogy.

  • #2
    Can you post the link for James Lick's mthap..the one you posted is not working?

    Comment


    • #3
      Originally posted by Sw33tdecent View Post
      Can you post the link for James Lick's mthap..the one you posted is not working?
      That's weird, it works for me, but here's the actual URL:
      http://vps1.jameslick.com/dna/mthap/

      Comment


      • #4
        23 and me search mtdna

        I've asked 23 and me to make the mtdna data more accessible. specifcially, if you go to raw data, browse, and click on mtdna, youll see the data is there starting with position 1... BUT at present there is no way to jump to a specific position of interst or even to jump to the last page. one would have to hit next next next over and over again. so i asked them if they would put in searchabilty so that one could say ok, search for 16126 and then that position would come up. or at least let one go to the last page...
        they said probably yes, theyve forwarded it to their enginerring team and it will take an unkown amt of time.

        Comment


        • #5
          Originally posted by penguin View Post
          I've asked 23 and me to make the mtdna data more accessible. specifcially, if you go to raw data, browse, and click on mtdna, youll see the data is there starting with position 1... BUT at present there is no way to jump to a specific position of interst or even to jump to the last page. one would have to hit next next next over and over again. so i asked them if they would put in searchabilty so that one could say ok, search for 16126 and then that position would come up. or at least let one go to the last page...
          they said probably yes, theyve forwarded it to their enginerring team and it will take an unkown amt of time.
          If you download your raw data, you can browse it any way you want.

          Comment


          • #6
            Originally posted by penguin View Post
            I've asked 23 and me to make the mtdna data more accessible. specifcially, if you go to raw data, browse, and click on mtdna, youll see the data is there starting with position 1... BUT at present there is no way to jump to a specific position of interst or even to jump to the last page. one would have to hit next next next over and over again. so i asked them if they would put in searchabilty so that one could say ok, search for 16126 and then that position would come up. or at least let one go to the last page...
            they said probably yes, theyve forwarded it to their enginerring team and it will take an unkown amt of time.
            You can jump to a page with the position of interest by modifying the URL. If you start with page 1, then hit next, you'll see a URL like this show up in your browser:

            https://www.23andme.com/you/explorer...T&pos_start=70

            You can then change the start position to say 16120.

            https://www.23andme.com/you/explorer...os_start=16120

            Comment


            • #7
              great idea on the url. I completely missed that! i'm
              surprised 23 and me didn't mention that when we discussed it. (on raw data, of course i've downloaded it. not in the mood tho to write code to easily browse what I want. promethease is fun for some suff but doexnt acdress everything).

              Comment


              • #8
                23 and me does not match ftdna and 23and me is wrong

                update. Very frustrating. I used Ann's suggestion which worked (thanks). But the information I got is wrong, not just missing. First the missing stuff was as expected. On my own account, which was on the chip before the current one, it was just missing info at the positions I wanted to doublecheck.

                but the frustarting part was on my brothers account. His is the new chip. I went to the three positions I wanted to check and his claims that he does NOT have the mutation at those three positions. these three mutations are not ones that define the haplogroup. For the ones that define the haplogroup, his and mine agree perfectly.

                I can see why 23 and me only sequences on the defining regions, but to claim they sequence in other regions only to have the data be wrong is very sloppy careless work and makes me rather suspicious.

                I know I do not have heteroplasmy at those sites (I had checked the pherograms myself) and there's no reason my brother's should be incorrect. While I was willing to entertain that one especially unusual one that I've wondered about could be different (a long shot certainly), all 3 would not be.
                incidentally, the 3 I checked were in the control region. in the coding region, I checked only 1 of 2 private mutations and they didn't sequence that position.

                so now what? i'm getting tired of emailing 23 and me. How much can I trust the rest of their data now?

                Comment


                • #9
                  Penguin, chip technology is especially difficult for mtDNA -- it's hard to design probes for closely spaced mutations. Version 3 introduced some experimental probes, and some of them are definitely not working as desired. If you run the raw data for you and your brother through James Lick's utility James Lick's utility, it will bypass the problematic ones.

                  http://vps1.jameslick.com/dna/mthap/

                  it's probably better to continue this thread elsewhere. You can start a thread at 23andMe or write to me privately [email protected].

                  Comment


                  • #10
                    Sounds to me like 23andMe owes you a re-test.

                    Comment


                    • #11
                      Originally posted by JOlson View Post
                      Sounds to me like 23andMe owes you a re-test.
                      Surely you jest? You of all people should be aware that there is an unavoidable error rate for chip technology. Bear in mind that there were 3 possible errors out of ~ 2700 mtDNA SNPs, and I wouldn't be surprised if some of those were in regions that are especially difficult for mtDNA, e.g. the poly-C region in the vicinity of 16189.

                      Comment


                      • #12
                        I uploaded my 23andMe results (Genome) and JamesLick.com told me that each of the subclades of T were not an exact match for me, and gave the reasons.

                        I uploaded my Fasta from FTDNA and it fit me exactly into T2.



                        Strange..

                        Comment


                        • #13
                          wrong - its a HUGE error

                          Originally posted by Ann Turner View Post
                          Surely you jest? You of all people should be aware that there is an unavoidable error rate for chip technology. Bear in mind that there were 3 possible errors out of ~ 2700 mtDNA SNPs, and I wouldn't be surprised if some of those were in regions that are especially difficult for mtDNA, e.g. the poly-C region in the vicinity of 16189.
                          Heavens no, you have the wrong interpreation - it's not 3 out of 2700 snps. I did not do a side by side comparison of 2700 snps. I went straight to my 3 private control region mutations and all 3 were shown as not present in my brother. The statistic of interest is how many private mutaions are they getting wrong. So far it's 3 out of 3. This is a HUGE error, not 3 out of 2700! I have 2 further private mutations in the coding region. as I mentioned, one they didn't seuqence, and the other I didn't look up. so so far, of private mutations in regions they sequence they have a 100% error rate. this error rate might be as low as 75% once I check that last mutation.

                          (of course they can get the non-mutated ones right! and of course they can get the non private mutations right that define haplogrops and subhaplogroups if they are specifically looking for those).

                          Your calculation is way off. 3 out of 3 is not an unavoidable error of the chip technology!

                          I have emailed them. let's see what they say.

                          (p.s. if there are certain probes that arent' working, they should say so. or say that raw data for anything but the defining snps should be ignored in mtDNA. Otherewise, we work from the assumption that the raw data is valid. Using someone elses 3rd party software like you suggest is not a subsitute for providing accuarte data or at least warning what is not accurate. it ruins research to have false data - and you if anyone should appreciate that.)
                          Last edited by penguin; 18th August 2011, 03:15 PM.

                          Comment


                          • #14
                            Originally posted by penguin View Post
                            Heavens no, you have the wrong interpreation - it's not 3 out of 2700 snps. I did not do a side by side comparison of 2700 snps. I went straight to my 3 private control region mutations and all 3 were shown as not present in my brother. The statistic of interest is how many private mutaions are they getting wrong. So far it's 3 out of 3. This is a HUGE error, not 3 out of 2700! I have 2 further private mutations in the coding region. as I mentioned, one they didn't seuqence, and the other I didn't look up. so so far, of private mutations in regions they sequence they have a 100% error rate. this error rate might be as low as 75% once I check that last mutation.

                            (of course they can get the non-mutated ones right! and of course they can get the non private mutations right that define haplogrops and subhaplogroups if they are specifically looking for those).

                            Your calculation is way off. 3 out of 3 is not an unavoidable error of the chip technology!

                            I have emailed them. let's see what they say.

                            (p.s. if there are certain probes that arent' working, they should say so. or say that raw data for anything but the defining snps should be ignored in mtDNA. Otherewise, we work from the assumption that the raw data is valid. Using someone elses 3rd party software like you suggest is not a subsitute for providing accuarte data or at least warning what is not accurate. it ruins research to have false data - and you if anyone should appreciate that.)
                            You still haven't sent me your raw data for analysis to see if the errors are in the experimental probes. I do know that some of those are definitely errors, and I have requested that they be blacklisted when they update the mtDNA analysis.

                            But as far as the chip is concerned, it's agnostic about whether a SNP is used to define a haplogroup. The probes are designed to detect a specific base (ancestral or derived) at a specific position.

                            Comment


                            • #15
                              Originally posted by Ann Turner View Post
                              Surely you jest? You of all people should be aware that there is an unavoidable error rate for chip technology. Bear in mind that there were 3 possible errors out of ~ 2700 mtDNA SNPs, and I wouldn't be surprised if some of those were in regions that are especially difficult for mtDNA, e.g. the poly-C region in the vicinity of 16189.
                              You're probably right. My comment WAS a bit harsh. I did an analysis a while back on autosomal results, and there is an amazing amount of variance in the number of no-calls between results. Probably due more to poor sampling technique rather than the lab work.

                              Comment

                              Working...
                              X