Announcement

Collapse
No announcement yet.

Too Much no-calls, bad quality! FF data should be retested.

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

  • Too Much no-calls, bad quality! FF data should be retested.

    Hello,

    I did FF testing in January 2015 but I found out now that the percentage of no/call is over 2,8%. So when I upload the data to Gedmatch or other sites I get a lot of misleading results, I have useless data.
    Meanwhile I already sent some new swabs, what should I do to be retested? Is there any way to ask to FTDNA to repeat my autosomal test? Frankly speaking I was surprised to discover such bad quality in testings from FTDNA.

    Thanks and kind regards.

  • #2
    I checked on no-calls for my recent kit using
    the gedmatch File Diagnostic Utility and got
    less than 0.6 % so yes 2.8 is a higher than expected
    percentage

    I would be interested in doing some analysis on your kit
    please send my your kit number via a private message.

    There are many reasons for a high % of no calls and the
    probably the biggest one is the quality/quantity of DNA
    you sent back to FTDNA

    So I would suggest that any retest be done with
    you giving a new DNA sample and you should make
    sure you follow the instructions
    The time and pressure of your scrape(brushing) of the cheek
    is the key. I always double the time they say and
    in all the kits I have done I have never had any trouble

    I am surprised that 2.8% would make a difference
    because there is a much larger difference than that
    in the SNPs from different companies.
    What does gedmatch say for your kit for the value of
    "Number of SNPs utilized by GEDmatch template"
    You can find that in the analyze utility
    What is it that makes you say the matches are unusable?

    Comment


    • #3
      Hello JSW,

      my kit number is E9560, in Gedmatch is FE9560.
      The GEDmatch DNA file diagnostic utility says the following:
      "No-calls = 20191 = 2.8779490746548 percent" and there are not any other major issues with the kit, so the data can be used and apparently is reliable. The problem arises when I use the "one to many" tool and I get a lot of false matches, it's a chaos.
      For instance if I use "People who match both kits, or 1 of 2 kits" tool I get some matches that after a triangulations seem false because they don't match one each other, it's useful to become crazy.
      Likely the same problems exists when I use the oracles, even if less evident, how precise can be the populations list?
      Another point is that the system impedes me to order again the same order for FF, I can't order twice the same test, even in case I decide to repeat it.
      Ok, I'm not an expert, suggestion or correction of what I'm saying are welcome of course.

      Thanks and kind regards.

      Comment


      • #4
        Can you please send me a private message with a good contact email address. If the no call is that high I would like to get some input from the lab or data assurance team. Thanks.

        -Darren
        Family Tree DNA

        Comment


        • #5
          Childs dna sometimes larger than fathers, EH?

          I also have a question related to the quality of test results:
          Both myself and my father have Family Finder results (tested a year apart) with many near identical matches as you would expect.
          However, for SOME of our common matches I have more shared cMs and longer blocks with them than what my dad does with them?
          How can this be, when my dna is supposed to be 50% more diluted than his is? Surely, my dad should have more cMs and longer blocks with them than what I do and not vice versa?
          Is there any logical scientific explanation for this phenomenon?

          Comment


          • #6
            Originally posted by paulanthony1976 View Post
            I also have a question related to the quality of test results:
            Both myself and my father have Family Finder results (tested a year apart) with many near identical matches as you would expect.
            However, for SOME of our common matches I have more shared cMs and longer blocks with them than what my dad does with them?
            How can this be, when my dna is supposed to be 50% more diluted than his is? Surely, my dad should have more cMs and longer blocks with them than what I do and not vice versa?
            Is there any logical scientific explanation for this phenomenon?
            Your parents are distantly related. I'm not talking ancient times, but within say the last ten generations.

            Jack Wyatt

            Comment


            • #7
              Originally posted by paulanthony1976 View Post
              I also have a question related to the quality of test results:
              Both myself and my father have Family Finder results (tested a year apart) with many near identical matches as you would expect.
              However, for SOME of our common matches I have more shared cMs and longer blocks with them than what my dad does with them?
              How can this be, when my dna is supposed to be 50% more diluted than his is? Surely, my dad should have more cMs and longer blocks with them than what I do and not vice versa?
              Is there any logical scientific explanation for this phenomenon?
              Although it could be that parents are distantly related, it is not as clear cut as that. No calls in fathers file may cause these blocks not to show up. I Think I read somewhere that FTDNA uses approximately blocks of 50 SNPs?? to match consecutively to form a segment of at least 1cM.

              Look at your fathers Raw data file at the tail ends of the segments (Start/stop)depending on which end has the shorter start or stop point.
              How many no calls are present in your fathers file on section that you match these matches but your father does not?
              Last edited by prairielad; 19 November 2015, 04:58 PM.

              Comment


              • #8
                Originally posted by prairielad View Post
                Although it could be that parents are distantly related, it is not as clear cut as that. No calls in fathers file may cause these blocks not to show up. I Think I read somewhere that FTDNA uses approximately blocks of 50 SNPs?? to match consecutively to form a segment of at least 1cM.

                Look at your fathers Raw data file at the tail ends of the segments (Start/stop)depending on which end has the shorter start or stop point.
                How many no calls are present in your fathers file on section that you match these matches but your father does not?
                Mr. Anthony had a thread a couple of weeks ago:

                http://forums.familytreedna.com/showthread.php?t=38736

                The scenario there also has the answer that his parents are distantly related.

                Jack Wyatt

                Comment


                • #9
                  Sorry to take so long to reply to this

                  Re "For instance if I use "People who match both kits, or 1 of 2 kits" tool I get some matches that after a triangulations seem false because they don't match one each other,"

                  I note that you have only one match with the largest segment
                  longer than 20 cM and that one is 28 cM then the next one is 15.6 cM. This tells me that all your matches are really
                  distant cousins probably back as many as 10 or more generations.
                  In general I do not chase matches below 20 cM as they are all lost in history..

                  You need to be aware of a couple of issues. One is that
                  as the segments get shorter the odds of the match being
                  a random (not real) match goes up.
                  The longest random match I have found is 14.7 cM
                  and the 50/50 point is about 7.8 cM at 10 cM it is about
                  20% random matches.

                  Also the odds of two 4th cousins actually having DNA from
                  their common one or two ancestors much less than 10%
                  see this page for much more and for links to more data
                  http://www.isogg.org/wiki/Autosomal_DNA_statistics

                  So if you match A and you match B and they are more than
                  4th cousins you should expect that they will NOT match
                  each other.
                  If the do happen to match each other that is a plus for
                  you and it means you might want to spend more time
                  on that match.

                  As to ancestral percentages - the statistics are such
                  that one cannot put a lot a faith in the results

                  An example from my family.
                  Three siblings with 3 of their grand parents are from colonial
                  America times and one is from London England with
                  unknown, but probably, deep English roots. I have them
                  back to the late 1700s.
                  When I look at the mix on their ancestors one would expect
                  to see about 25% British Isle and that is the case for
                  two of them but the 3rd one comes in at 0% British Isles.

                  So if the test cannot find 25% of one large population
                  how good will it be on any smaller one??
                  And having such different results for siblings tells me the data from any one person is highly suspect.
                  From an atDNA point of view it is difficult to sort out
                  English from German and from other western European populations.
                  On the other hand it is easy to sort English from Asian or
                  English from African.

                  Bottom line I think that while you have a larger than
                  average percent of no-calls I do not think it is
                  affecting your results

                  If you wish to get rid of some of the random matches
                  you can try using phased data. Phased data will cut the
                  random match points in half. So 7.5 cM becomes good
                  and the 50% point is at 4 cM.
                  If you have a living parent you can use their data to phase yours.
                  If you have children you can use yours to phase their's
                  and from that deduce the status of a segment match to you
                  as to it being real or random.
                  Good luck on your search
                  John W

                  Comment


                  • #10
                    Originally posted by JSW View Post
                    In general I do not chase matches below 20 cM as they are all lost in history.
                    John W
                    If I did that, here at FTDNA I would have only about 25 matches in all for all my family members combined!!!

                    Comment


                    • #11
                      Re number of matches greater than 20 cM

                      On gedmatch.com I have 129 matches greater than 20 cM
                      and if I included 17 cM and up it is 258

                      I guess it all depends on the population tested and
                      since am am mostly of Colonial American I have more than
                      some others.

                      John W

                      Comment


                      • #12
                        My mother has only one grandparent who had Colonial American Ancestry though, so I have to use smaller segments. Her siblings are deceased, so I have to use my cousins' results. They have even smaller segments and very few matches.

                        As for the French and German immigrants the surprising thing is that she has a Harpring and a Steffen match here. At Ancestry she has no Harpring, Steffen, Buttermann, Gohmann, Hartkemeyer, Kaisekamp, Elter, Villain, or Reckendrees, etc. matches. She has matches there with surnames in her Messmer line who have ancestors in the Baden area. I suspect they are related to her ancestors in the area around Surbourg, France who migrated there from Switzerland after the 30 Years War. I don't think we'd ever find the connection though, and that's one branch of my tree that I've been able to trace back to the 1600s through church records.

                        Comment


                        • #13
                          No Calls

                          Does anyone know what an acceptable number of no calls is?
                          At what point should we raise the red flag with FTDNA?

                          Comment


                          • #14
                            Originally posted by Grapeleaves View Post
                            Does anyone know what an acceptable number of no calls is?
                            At what point should we raise the red flag with FTDNA?
                            Well, I don't know what the threshold should be, what I can say for sure is that my percentage of no-call is very close to 3% while usually other kits do not reach 1%.
                            I sent twice my swabs to FTDNA because while doing FF they warned me that the sample was too old, so they already have a new/fresh sample available, but they used the old one, I'm wondering why...I wasted even other money to send again the swabs that then they ignored.
                            Last but not least they asked me to pay again to retest the FF if I want to try to have a lower level of no-calls, at my own risk and without any assurance on the quality of the results, 3% is really too high.
                            In few words it seems their threshold are comfortable only for their labs (..) but not for us if we want to export data to use it elsewhere.
                            The result is that if I use my data in gedmatch I get a lot of false matches and even the oracles likely return a lot of unreliable data.
                            Frankly speaking now I'd prefer to abandon all the matter and to spend my money in a good restaurant, it's more healthy.

                            Comment


                            • #15
                              Hi Minnose,

                              Was this a recent attempt to resolve the issue? I never received a PM from you so I am not certain if you contacted the helpdesk independently more recently. If you would like to send me a Private Message I would be happy to at least look into the issue.

                              -Darren
                              Family Tree DNA

                              Comment

                              Working...
                              X