Announcement

Collapse
No announcement yet.

23andme v4 Transfers - What's the official word

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

  • #31
    It's clear that v4 is not compatible with FTDNA's transfer service, but will it ever be? Or is the data inherently something that FTDNA's systems could never work with?

    On a tangent, has there been any question on v4's accuracy in terms of Ancestry Composition vs. V3?

    Thank you.

    Comment


    • #32
      Originally posted by Nijuurasen View Post
      It's clear that v4 is not compatible with FTDNA's transfer service, but will it ever be? Or is the data inherently something that FTDNA's systems could never work with?

      On a tangent, has there been any question on v4's accuracy in terms of Ancestry Composition vs. V3?

      Thank you.
      I think it's more a matter of whether the IT department has the wherewithal to investigate the possibility and adjust their methods to allow a smaller number of SNPs, but we need some empirical data. They could simulate the consequences by substituting no-calls for the missing SNPs in some of their existing records and see how much that changes things.

      I have v4 data that I can compare with my old results. My match lists are virtually identical, but average length of the segment is about 0.1 cM longer. This is because the v4 chip has fewer SNPs, and it takes a tad longer to encounter a contradiction to the long consecutive run of half-identical SNPs. The net result is that my v4 self has one more match.

      GEDMatch can certainly handle v4 data OK. John Olson wrote "the results are roughly comparable... except maybe at the very low end where matches would be marginal in any case."

      Comment


      • #33
        Thanks for that.
        Yes, I'm using both v3 and v4 data on gedmatch, seemingly without problems.
        More than matches I was curious about ancestry results (ancestry composition, gedmatch oracles) being less accurate with v4. Not the case, I guess.

        Originally posted by Ann Turner View Post
        I think it's more a matter of whether the IT department has the wherewithal to investigate the possibility and adjust their methods to allow a smaller number of SNPs, but we need some empirical data. They could simulate the consequences by substituting no-calls for the missing SNPs in some of their existing records and see how much that changes things.

        I have v4 data that I can compare with my old results. My match lists are virtually identical, but average length of the segment is about 0.1 cM longer. This is because the v4 chip has fewer SNPs, and it takes a tad longer to encounter a contradiction to the long consecutive run of half-identical SNPs. The net result is that my v4 self has one more match.

        GEDMatch can certainly handle v4 data OK. John Olson wrote "the results are roughly comparable... except maybe at the very low end where matches would be marginal in any case."

        Comment


        • #34
          Just bumping this thread to see if there's been any development on 23andme v4 autosomal transfers being accepted at FTDNA.

          Comment


          • #35
            Is any news?

            Comment


            • #36
              V4 chip transfer

              Are there any plans to modify programming to accept? Unfortunately, I have a family member who I don't think is willing to test a 2nd time and is elderly with health issues. If FTDNA never accepts the V4 transfer, countless people will never benefit from being on your family finder listing.

              Sue

              Comment


              • #37
                I don't know why 23&Me went to v4 chip, save money and stop clients from transferring probably. In short looks like too much guess work has to go into matching v3 & v4 chip data.

                Felix has a couple of blogs that compare the V4 chip here is one http://www.fi.id.au/2015/01/match-di...a-testing.html

                Conclusion
                FTDNA and Ancestry kits are perfectly compatible with each other. They are also compatible with 23andMe v3. However, 23andMe v4 kits are not compatible with FTDNA, Ancestry or 23andMe v3. They seem to drop off segments when compared with 23andMe v4 and creating many differences in segment lengths and total segments. Also, 23andMe v4 kit comparing with 23andMe v4 kits seems to add segments not found when compared with FTDNA and Ancestry. Caution must be taken when taking 23andMe V4 matches as it does not always go well with FTDNA and Ancestry kit matches.

                Comment


                • #38
                  All the companies are now using a "quad" chip, which lets them run more samples at a time. That does reduce the cost of a test.

                  However, a quad chip wouldn't have enough room for the standard 700K SNPs plus all the custom SNPs that 23andMe includes for medical, mapping, and ancestry purposes (Y and mtDNA). Thus 23andMe's v4 chip is completely customized for them, but they retained enough SNPs to be backward compatible and useful for ancestry within their own database. I have compared my v4 test with a v2 test, and I get pretty much the same matches, although the length of a segment averages about 0.1 cM longer.

                  It's not 23andMe's responsibility to make their chip compatible with other companies, but their motivation has nothing to do with preventing transfers, either.

                  I happen to believe that there ARE sufficient SNPs to be useful in cross-platform comparisons, but the algorithms would need to be adjusted. I have looked at some of Felix's examples, and the main difficulty seems to be the requirement for an arbitrary number of SNPs, even within a large cM segment.

                  This can be adjusted by the user at GEDMatch. However, FTDNA includes small segments in their algorithm, and the reduced number of SNPs does present an obstacle for FTDNA's algorithm.

                  FTDNA's algorithm also creates false negatives when people don't have enough small segments to meet their threshold. For example, many people have reported that about 20% of a child's matches aren't found in either parent. Some of those will be false positives in a child, but some of those will be false negatives in the parent.

                  Disclosure: I have had consulting agreements with 23andMe.

                  Comment


                  • #39
                    Is any news? Would like to move all my family members to FTDNA.

                    Comment


                    • #40
                      Yes, any hope for 23andme v4 users

                      Just bumping this question. I have my Dad's Y-DNA on here. I tested on 23andme for the health info, but would like to have all the DNA results here. Thanks.

                      Originally posted by Wonderfalls View Post
                      Is any news? Would like to move all my family members to FTDNA.

                      Comment


                      • #41
                        Is there any word on when FTDNA will accept V4 transfers from 23andMe?

                        Comment


                        • #42
                          Transfers? From where??

                          So, FTDNA offers transfers for $39, but NOT from 23andMe and NOT from Ancestry... Where from, then?

                          Comment


                          • #43
                            Transfers are possible from Ancestry and 23 and Me but only up to a certain date when these companies changed their testing techniques.

                            Comment


                            • #44
                              You can transfer any Ancestry test taken before May 2016 and any 23andme test taken before Dec 2013 (believe that is the date 23andmev4 came into effect)

                              FTDNA tests around 716,000 SNP including 17,800 X SNP

                              SNP tested on old Ancestry chip shared with FTDNA approximately 689,000 SNP including 17,400 Shared X SNP

                              SNP tested on new Ancestry chip shares with FTDNA only approximately 427,000 SNP including 16,400 Shared X SNP

                              SNP tested on old 23andme chip (V3) shared with FTDNA approximately 690,000 SNP including 17,100 shared X SNP

                              SNP tested on new 23andme chip (V4) shares 312,000 SNP including only 7,500 X SNP

                              Comment


                              • #45
                                @prairielad

                                Very interesting information, do you perhaps have a source about it?

                                Comment

                                Working...
                                X