Announcement

Collapse
No announcement yet.

Help! Gedmatch upload - No-calls = 31101

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

  • #16
    Originally posted by prairielad View Post
    [...]
    Gedmatch only extracts from your Raw Data file the SNPs which they have in their template
    So did GEDmatch change their template recently, so that FTDNA data just looks like it more no calls than it used to in prior years?

    Comment


    • #17
      My take on the ~30K no-calls is that these are medically-significant SNPs that the chip reports but that FTDNA does not use because they don't want to be in the medical diagnostics business. Rather than just deleting them, they report them as no-calls.

      Comment


      • #18
        Originally posted by rd298935 View Post
        So did GEDmatch change their template recently, so that FTDNA data just looks like it more no calls than it used to in prior years?
        I don't know exactly what constitutes the "template" at GEDmatch, but I could see the no-calls in the raw data you sent me.

        Comment


        • #19
          Originally posted by DRNewcomb View Post
          My take on the ~30K no-calls is that these are medically-significant SNPs that the chip reports but that FTDNA does not use because they don't want to be in the medical diagnostics business. Rather than just deleting them, they report them as no-calls.
          I rather doubt that explains the current scenario, with a large jump in the number of no-calls. FTDNA has redacted about 3,000 "medical" SNPs from the very beginning, and those don't show up in raw data downloads at all. I deduced this by comparing the raw data download for an AncestryDNA v1 upload.

          If I had to guess, I'd say Illumina has been supplying chips with a reduced number of SNPs. This happened once several years ago.

          Comment


          • #20
            Originally posted by rd298935 View Post
            Meanwhile, my support ticket got the reply that the two kits percentages (4%) are within their "quality control checks and measures."

            And that:
            "Some atDNA SNPs which are not included in our test have placeholders in the raw data file for formatting purposes which can inflate a different service's no call calculation."

            I'm going to ask what that means. In other words, formatted for who?
            Presumably formatted for themselves. They certainly aren't catering to GEDmatch. FTDNA's calculations may be simpler if everyone has something reported for every SNP, so the 17856th row of data is the same SNP for everyone in their database.

            Comment


            • #21
              As GEDmatch is in the unique position of comparing SNP's from a large and growing number of different vendors, it would make sense to maintain a "template" or standard list of SNP's that have characteristics suitable for genealogical use. Among other reasons, we would not want to include SNP's that occur (in their variant state) so seldom that they would match in virtually everyone who was tested. A large number of such SNP's could result in spurious matches. In fact, there was a rash of absurdly numerous close matches some months ago, apparently due to one or more vendors "imputing" a large number of SNP's. GEDmatch has apparently found a way to eliminate these situations. A "template" that specifies the SNP's that are acceptable for use in the matching algorithm would be a good way to accomplish that goal.

              Comment


              • #22
                By leaving no-data placeholders for such a large number of SNP's, FTDNA has made it impossible to assess the quality of the raw data. The number of no-calls is now essentially meaningless. Not a good strategy for the future!

                Comment


                • #23
                  same result for mapmy23

                  [QUOTE=John McCoy;453162]Pretty sure "no-calls" are not counted as mismatches by GEDmatch, so the effect should be minimal,]


                  I had to use the mapmy23 tool in order to upload my RAW Ancestry DNA file to FTDNA and when comparing the original with the modified version at GEDMatch I found the same thing had happened. The original Ancestry RAW file had only 353 no-calls according to GEDMatch's file diagnostic utility but RAW DNA data as modified for FTDNA by mapmy23 came up with 33,700 no-calls, or about 7.5%. There were actually fewer matches for the modified data with the higher number of no-calls; 420 fewer in the >=7cm range and about 3000 fewer matches for the remainder.

                  That said, when spot-checking the '1 to many' (>=7cm) matches for each upload side by side in spreadsheets the names listed for either kit were pretty much the same.

                  OTOH, FTDNA is only showing me 54 matches where Ancestry, GEDMatch and MyHeritage each show thousands.

                  Comment


                  • #24
                    OTOH, FTDNA is only showing me 54 matches where Ancestry, GEDMatch and MyHeritage each show thousands.
                    You might want to check your account settings, under "Privacy & Sharing." Make sure your matching preferences are set to the level you want, such as "All Levels" or "Distant, Close & Immediate." If they are set to "Immediate Only" or "Close & Immediate," you will not see very many matches. That would explain the difference in the quantity of your matches at FTDNA vs. the other companies/places. Make sure the blue "Consent to Participate in Matching" statement has been checked, also.

                    Comment


                    • #25
                      [QUOTE=Ralf_G;453486]
                      Originally posted by John McCoy View Post
                      Pretty sure "no-calls" are not counted as mismatches by GEDmatch, so the effect should be minimal,]


                      I had to use the mapmy23 tool in order to upload my RAW Ancestry DNA file to FTDNA and when comparing the original with the modified version at GEDMatch I found the same thing had happened. The original Ancestry RAW file had only 353 no-calls according to GEDMatch's file diagnostic utility but RAW DNA data as modified for FTDNA by mapmy23 came up with 33,700 no-calls, or about 7.5%. There were actually fewer matches for the modified data with the higher number of no-calls; 420 fewer in the >=7cm range and about 3000 fewer matches for the remainder.

                      That said, when spot-checking the '1 to many' (>=7cm) matches for each upload side by side in spreadsheets the names listed for either kit were pretty much the same.

                      OTOH, FTDNA is only showing me 54 matches where Ancestry, GEDMatch and MyHeritage each show thousands.
                      You shouldn't use the converted file from mapmy23 for other sites like GEDMATCH, its only meant for FTDNA in the format they accept(I'm the author). GEDMATCH can accept pretty much any format so there is no point converting it and discarding the data that ftdna doesnt use.

                      Comment

                      Working...
                      X