Are you confused about using auDNA on Wikitree?

+11 votes
949 views

Wikitree users are confused as to the definitions of “father is confirmed with DNA” and “mother is confirmed with DNA”.  Wikitree DNA users are confused as to what rules to apply in different scenarios.

Here’s one way to know if you are confused.  If you believe that the choice of “father is confirmed with DNA” means that DNA confirms that the person identified in the Father profile is the father, then you are confused.  DNA does not make such assertion; it only asserts that the DNA further supports the Father identified is the father. It does not rule out other possibilites.

I apologize for being blunt, but Wikitree leadership and the members of the DNA project are confused as to the difference between “Supporting a parent/child relationship” and “Supporting all other relationships”, including cousins. There is a significant difference in the two.

This difference corresponds to the guidelines used by genetic genealogists for matches with documentation (relationship is known), and a different set of guidelines for matches without documentation (relationship is not known).

The objective expressed by Wikitree ONLY applies to that part of Genetic Genealogy that relates to matches with Documentation; the relationship is known. This part is NOT the part of Genetic Genealogy that relates to matches without Documentation; the relationship is NOT known.

Genetic Genealogists who map genomes require known cousins whose DNA Prediction is consistent with being a known cousin.  Identifying profiles to Genetic Geologists that can be mapped is the Wikitree added value to these Genetic Genealogists. On Wikitree, DNA should be used to identify those profiles in which a Non-Parental Event is unlikely.  A Non-Paternal Event is unlikely if we know the profile can be mapped.

You do not need to look at the segments to complete the task of determining what profiles, in a parent/child relationship are mapping candidates. Any Genealogist of any technical level can use any DNA Service’s predicted relationship, for every match. The current guidelines, IMO, exclude 95% of the genealogists, and 95% of the available matches for no scientific reason.

This brings me to the part of Genetic Genealogy that relates to matches without Documentation.  Unlike the first part, which does not require knowledge of segments, this second part does require them. This second part is not interested in the parent/child relationship but is interested in identifying common ancestors with cousins with documentation. In other words, they want to prove a theory of the xCousin relationship.

The best approach for finding the documentation for those DNA testers without, is to identify, by the using of matching and triangulation, the most likely common ancestor(s).  The many articles that have been used to argue for and against which segments should be considered evidence ONLY applies to that part of genealogy related to matches without documentation.  This is not part of the defined scope articulated by Wikitree.

Here is an example of such an article.

http://www.yourgeneticgenealogist.com/2014/12/the-folly-of-using-small-segments-as.html

in general, people tend to see evidence that supports their theories”

Theories are those instances when there is no documentation; this approach is not intended to be used by those instances in which their already is documentation to support a claim.

in Policy and Style by Ken Sargent G2G6 Mach 6 (61.6k points)
edited by Ken Sargent
As far as predicted relationships go- here is a good example of why you can't rely on the company for accurate relationship predictions:

My top 5 matches, all known, all well documented, are all listed as 1st cousin. NOT ONE of these 5 individuals are my first cousin. 1 is my mother's fraternal twin sister, one is her older sister, one is my father's sister. In other words, 3 full aunts. 1 is my maternal half brother, and the last is my paternal half sister. These are as immediate of a relationship as I can obtain from living humans. All predicted incorrrectly. But I DONT need to look at their segments to prove a relationship because the amount of shared cM's is what confirms this, not what location on what strands. Really, its more of a hassle to list all 64 segments and their locations on wikitree than anyone has time for.

Why would I bother looking at shared segments for this close of a relationship? To establish general guidelines for every match that is a relationship further removed. Instead of spending months trying to figure out who our MRCA is, using a family tree as our tool, or my favorite, comparimg our ancient ethnicity as a means of how we might be related, a quick glance at location can use the process of elimination or simply refer to known segments to tell you where to look for a probable MRCA. But from the few confirmed DNA relationships I have done on wikitree, if the relationship is 3rd cousin or closer, you do not need to specify what segments you match on, and you do not need to provide the info of any 3rd parties for triangulated proof. They only ask that you provide the shared amount of cM's that confirms or supports the relationship YOU are claiming with your match on THIS SITE, which does not provide you with any algorithms to prove it. If I say my 1/2 sister is my 1/2 sister because ancestry says so, are you going to believe me? Even with the claimed amount of cM's from ancestry, which is different on GEDmatch for every person I've compared with at both sites, BTW, and their reliable relationship prediction of my sister being my 1st cousin? Maybe. The more info I give, the more it helps for you to trust that I know what I'm doing. What if I think I know everything? Like most people in the family tree community? You know, THOSE kind, who tell you that you have an ancestor wrong and demand you adjust your tree immediately, but don't say sorry when you slam them with documents to the contrary and DNA evidence? What if I was a DNA tested "Tree Meddler"? And because I always know everything, tell you that I've confirmed a 2nd cousin relationship on ancestrydna because we share 26cMs? Well, Peter will catch it, and you know that the relationship is either NOT correct based on that small of shared DNA, OR it IS correct, but you got the luck of the DNA dice where the average amount of DNA handed down from parent to child has been tipped to one side a bit more. I am one of those people who just didn't get ANY DNA from a branch of the family! I match nobody fthat is a relationship of 3rd cousin on that line to my mother via DNA...Without my closest relatives all testing, I would have used science to determine that there was a mistake in my tree! In those cases, it's for the sake of the potential cousin match that I would include more info explaining in detail how I decided that relationship was proven. This also goes for posting my known segments. This way, you don't have to contact the distant stranger cousin. Someone has given you a ton of work they did to help YOU not have to do it too, it's a courtesy.

Now. I also have the same set of grandparents on both sides of my tree. Further back, way back, and on dad side, they are 6th or 7th great grands, and on mom's side, 5th greats. When I first got my test results from ancestry, I didn't have them yet in my tree on dad's side, only mom's. The predicted relationship of anyone who matches me at that distance is in the grey area of shared cMs. I cross checked my aunt and sister on dad's side to see if they shared DNA with this match, they didn't, so I was excited to confirm this relationship. Then, my brother and 1c1r showed up, and I made a breakthrough in dad's tree.... turns out, not a maternal match after all. My sister and aunt just didn't get that part of the paternal shared DNA. How would I determine every match from here on out with that set  of grandparents? Take it to GEDmatch.

What about the one lady who matches me on ancestry and GEDmatch? Jan? She is a nightmare for me! Ancestry gave me a "hint" that was soooo wrong! They said we shared a set of paternal grandparents, which is true! Gold money! But... no. Wait. She matches my brother and aunt on mom's side! So then, revamp to looking at names. Must be Conway- but then, wait! Mom's first cousin John's test results pop up, and who's on the list? AAAAARGH! Jan! So, not mom's paternal line, but maternal. But WHO? Where? I don't know, nobody does, but if I can figure it out, I'm telling EVERYONE! None of our shared matches are giving up any secrets with their locked profiles and lack of trees either. This is going on almost 2 years! And when I tell everyone, I most certainly will go above and beyond the set requirements for relationship verifications per wikitree' guidelines and put too much info for the next person who has Jan. That algorithm pattern used by my testing company truly outdid itself that time!

All this being said because you CANT just sit back and let the algorithms give you a family. However much money and people and time they put into your report, it isn't PROOF. Had I not have known what little I know, I would be so screwed in my personal perceived knowledge of my matches being confirmation of anything at all if the Big Bang corporate test provider was my only method of deciding who stays and who goes! I also can thank my testing company for teaching me some lessons about how DNA works, because without their many many mistakes, I would have trusted them with their reports.

I'm against the whole "confirmed relationship" via DNA thing on this site anyway. It's not wikitree who is confused, it's everyone who uses the site. It's threads like this that cause one to just say "F*** it" and ignore the DNA button. Simply providing your kit for comparison or letting folks know you tested, add that  with a tree is a positive way to avoid the repetition of explaining confirmed relationships and STILL nobody gets it... love this site though, and recommend to everyone!

6 Answers

+6 votes

Please point out an example where WikiTree is using DNA for "supporting all other relationships, including cousins." Thanks.

by Peter Roberts G2G6 Pilot (695k points)
Well it does create back to 6ggparents and out to 3rd cousins, but that is just to indicate there might be a match, so other relatives can compare their DNA with say, mine and see what it shows.  So far I haven't had any write to me that they see a match, but I suspect it will happen one of these days.
Peter, I never said Wikitree is using DNA for "supporting all other relationship, including cousins."  I said Wikitree is confused. The current guidelines, along with numerous responses, indicate that they are using the process used in Genetic Genealogy to prove a cousin relationship, which is not necessary, to "support a parent/child relationship."

If I have a documented relationship with a cousin and the match is consistent with that documented relationship, this is enough to map profiles.  If Genetic Genealogists can map a profile, by extension, there is a significant unlikeliness that a NonParental Event has occurred.

Why does wikitree use the guidelines used by Genetic Genealogists to identify the cousin relationships that have no documentation?  In other words, Wikitree requires that DNA proves the Relationship of the match.

Every response is aimed at what proves a 4th cousin is a 4th cousin, but mapping  would only care about establishing a relationship that establishes the match is at least a 1/2 4th cousin, and it doesn't care if the match is a full cousin or which is the Most Recent Common Ancestor.
+4 votes
Well, I know a good bit about genetics and DNA, but I can't say I know exactly how the genetic DNA tests are designed.  Still, I haven't particularly noticed any touting on WikiTree of being able to identify relatives exactly without documentation.  It may be that some individual WikiTreers get things wrong.  After all it is a cooperative site and not everyone will draw the same conclusions from a given set of data.

BTW, I have grave doubts about that article you link.  At least two sentences I saw in a fairly fast rreadthrough, seem to be definitely wrong, and in general he relies on particular cases of marginal segments to claim that such segments shouldn't be used at all.  In DNA genetics we're always making statistical comparisons.  The fact is that if we find a match at the 95% probability, this meas there will be 5% will be incorrect matches.  So you need a large universe of matches to draw a trustable conclusion.  But I don't see an attempt to provide us in this article a statistical argument with proof.  Anyway, the matches we see from say Ancestry.com don't rely primarily  on the <5cm segments but on much larger ones.  I did create some comparisons on GEDmatch and there was no sign of random large segment matches.
by Dave Dardinger G2G6 Pilot (438k points)

Dave, I agree with what you have stated. "Still, I haven't particularly noticed any touting on WikiTree of being able to identify relatives exactly without documentation"

My point is that the guidelines used are the same as the ones used to identify relatives without documentation. 

My point is why is this required when it is not necessary?

Dave, If you have a documented 3rd Cousin 1x removed, and 3 DNA Services and Gedmatch predicting a consistent relationship like predicted 3rd cousin in the range of 3rd to 4th cousin.  The is sufficient for a Genetic Genealogist to use this part of the tree to map.  It also tells a wikitree user that a Non-Parental Event has most likely not occurred.  This is the best auDNA can tell us, but Wikitree requires significantly more, but can only reach the same conclusion.
Another point,

If we have documentation that proves a 3rd Cousin 1x removed and DNA supports this, why do we need to look at segments and debate the conclusion? These DNA services have invested millions in gathering data and creating algorithms to make their predictions.  

Do we really believe we can do a better job looking at the segments and debating what is and is not relevant?
Well, can you point me to a WikiTree page that illustrates what you're saying?  Now, I think wanting to verify a relationship without using documentation can be useful.  This means that the documentation is also verified as pertaining to the person the test leads to.  But as you go back in time, there is the possibity of lots of possibilities.  In my tree, for example both Minerva Miller and her older sister Charlotte were married to Elijah Green.  Some of the documentation shows the child Minerva as being Charlotte's (i.e. they married the child after her late mother.  But other documentation gives the daughter a birth date before her mother died and before Charlotte married Elijah.  Now Matilda and Elizah had a previous child who may have descendents still alive, and there are definitely children of the child Matilda and of Elizah and Charlotte, so presumably this could be proven by comparing segments of child Matilda with those of adult Matilda and adult Charlotte.  Some segments would be lacking in both of the sisters and their occurance in the child Matilda would prove who her mother was.  But it's not high on my list of things to do since I trust the dates of child Matildas birth and mother Matilda's death (about a 2 month difference.  But you could come up with similar scenarios where you would need the DNA test to sort out potential foul play or hanky-panky.

Dave - 

I'm not sure which statements in CeCe Moore's article have errors (in your opinion).  But - if you want to see a larger study showing that using small segments in DNA genealogy is not reliable, here is another article from CeCe's colleague Blaine Bettinger which draws the same conclusion against the use of small segments from a study of 25,000 tests (which included 3,000+ tests from a parent-child)

http://thegeneticgenealogist.com/2014/12/02/small-matching-segments-friend-foe/

"In a study earlier this year using real genomes (including thousands of genetic genealogists) scientists from 23andMe examined IBD segments in a group of 25,000 individuals which included 3,000 father-mother-child triads.

In the study, a segment of DNA shared by a child with someone in the database other than their parents was declared to be IBD if that same segment of DNA was shared with a parent (importantly, the IBD segments only had to overlap by 80%, not a full 100%)."

"The researchers found that more than 67% of all reported segments shorter than 4 cM are false-positive segments (see FIG. 2B, below). At least 60% of 4cM segments were false-positive, and at least 33% of 5 cM segments were false-positive. The number of false-positives decreased fairly rapidly above 5 cM.

To my knowledge, this study is the largest of its kind to examine this question directly. The paper is available online for free (http://mbe.oxfordjournals.org/content/31/8/2212)."

So - whether we are talking about some anecdotal examples from CeCe's article or the larger study cited in Blaine's article, there is strong evidence suggesting against relying on segments smaller than 5 cM. 

Dave," Now, I think wanting to verify a relationship without using documentation can be useful. "

I had to prove a nonpaternal event and identify the actual biological father. In this case, it was close enough to verify without documentation but it took quite a few more tests than the guidelines state to do this.

The major group that uses segments to identify a common ancestor are Adoptees. In these cases, the best they can do is to use to look at the segments of known relatives and using triangulation, guess at their relationship with those 2 cousins.  When you know your relationship, you don't need triangulation. When you don't know your relationship, you do need segments and triangulation.  Wikitree only asks you to mark profiles that has documentation, so there is no need to look at segments or require triangulation.

 

Ray, 

I completely agree with the "larger study showing that using small segments in DNA genealogy is not reliable"

My question is:  "Why look at the segments at all?"  If you match a 3rd cousin 1x removed, and the DNA Service predicts you are 3rd Cousin in the range of 3rd to 4th cousin. We know the prediction service has invested millions in reaching this conclusion. We also don't know exact what segments the used or what they weight the carried. 

These articles are directed at those who have a predicted relationship but do not know how they are related, like adoptees.  These adoptees have to look closely at the segments and use triangulation to narrow down how they are related to a match.  These are not scenarios within the scope of the wikitree. 

I use segment matching regularly - often in conjunction with my paper research.

Here is an example of this (sorry if this is old, as I have mentioned it before on g2g).  My great-great-great-grandfather, John J. Coates (1815-1867) lived in Meigs County, Ohio.  A man named Jeremiah Coates (1817-1893) lived in Gallia County, Ohio.  Meigs and Gallia are adjacent on the map, but based on all of the paper records for John and Jeremiah and their respective issue, there is no evidence demonstrating that John and Jeremiah were related.

Autosomal DNA tests (all of which have been submitted to Gedmatch) on the descendants of John and Jeremiah provide the first evidence of a possible relationship.  We have tests from 2 of Jeremiah's descendants and 8 of John's descendants.  Since we can see the data on Gedmatch, we have been able to identify 10 different DNA segments that are between 7.3 cM - 33.5 cM in common between the descendants of John and Jeremiah.

Taking a deeper look, two of the segments achieve the standard of "triangulation" as they are shared among 1 of Jeremiah's descendants and 2 of John's descendants (making sure that the descendants of John are further apart than 1st cousins).

So - moving forward - as we get additional descendants of Jeremiah's and John's to test and submit their test results to Gedmatch, we already have 10 specific segments to examine for further evidence of triangulation - as well as any additional segments that might materialize from the new tests.

I'm not using anything smaller than 7.3 cM right now, but am encouraged to see that of the two descendants of Jeremiah, one has matching segments in common with 5 of the 8 descendants of John and the other has matching segments in common with 6 of the 8 descendants of John.

In this case anyway, looking at the segments in common among the tests is the critical issue.

Ray – I don’t know if this makes any difference, but I do know the value of using segments when working with DNA Relatives that you have not yet determined how you are connected. A few years ago, I built a backend database and began getting reports from RAW Data of my different family members.   I stopped working on that project and used my expertise in Database applications to help Gedmatch get back online uploading new kits.  Gedmatch connected me with experts in this area but a few weeks later, I had to stop because of health reasons.

I tell you this because when I say I completely agree with you as to the usefulness of segments in finding how we are connected to our DNA Relatives; I say this fully understanding the Science behind my statement. 

I contend that you are confusing using segments when the exact relationship is unknown with the Science behind further confirming with DNA the parent/child relationship where the parent and the child may or may not be a DNA Tester.

The Predictive Science used by the DNA Services does further support a parent is confirmed with DNA solely based on the DNA Service Prediction. 

Let me ask you questions another way.

Why do you think that any Wikitree Genealogist, with varying degrees of understanding DNA, will provide a better prediction by interpreting segments themselves than AncestryDNA, FTDNA, 23andme, or Gedmatch?

Why exclude the vast majority of genealogists from participating because of a lack of understanding or interest in segments, especially when they are not necessary.

More important, why exclude from further confirming with DNA the millions of DNA tests that are currently not uploaded to Gedmatch, especially since Gedmatch is not needed?

A known 3rd cousin who is predicted to be a third cousin clearly supports most of the claims of a father/mother is confirmed with DNA.  You may get a different relationship prediction when you make the prediction than those of the DNA Services, but which is authoritative outside of Wikitree?

I feel confident that a lightbulb moment will happen soon.

 

On the smaller segments discussion (which is different than the original post) I just discovered yesterday that two triangulated groups that my partner has at chromosome 16 (one is 3 members strong, the other one is 7 strong) at the same position (!) is perfectly ok when measured with a threshold down to 5 cM and 500 SNPs.

But as I lowered it down to 1 cM and 500 SNP's (as it's still 500 markers no matter if it's on a shorter string of DNA or not) it turned out to be that it's not two different TG's, it's actually one!

So much about shorter segments not worth it. You can't rule out shorter segments when you have paper trail proven cousins and when you have identified TG's through the general 700 SNP/7 cM (or 500 SNP/5 cM) threshold. Many missing matches between members of a TG are actually below the threshold as I found out through my own research on now more than 30000 matching segments.
+6 votes

I'm not following this point:

"Here’s one way to know if you are confused.  If you believe that the choice of “father is confirmed with DNA” means that DNA confirms that the person identified in the Father profile is the father, then you are confused.  DNA does not make such assertion; it only asserts that the DNA further supports the Father identified is the father. It does not rule out other possibilites."

When I do a one-to-one comparison on Gedmatch between my test and my Dad's test on Gedmatch, I see that we have 3586.9 cM in common.  In addition to this amount shared, the chromosome browser shows that every segment we share is identical.

Just on shared cM alone, this is scientific confirmation that my Dad is my father.  The only other possibility for people sharing this much DNA in common is an identical twin, which is definitely not a possibility for me and my Dad due to our age difference.  The 3,500+ shared cM between he and I makes it pointless to search for other possibilities as my father.  While this is a non-issue in my case, it could be a very significant point for an adoptee or a person with an NPE who finds a similar match.

So - when I list he and I as parent-child "DNA confirmed", yes the DNA test confirms that he is my father. 

by Ray Jones G2G6 Pilot (161k points)
Ray, why do you need to look at the segments in this case.  If AncestryDNA, FTDNA, and 23andme all predict a father/child relationship. Why would it be necessary to look at segments?

The parent/child relationship in your case, and the parent/child relationship between a great-grandparent and great-great-grandparent is still a parent/child relationship.  Why the need to look at segments?
In the case of a parent-child, looking at the segments on the chromosome browser shows you that every segment in common is fully-identical, which is further support of the relationship.  I agree though - with 3,500 cM shared, the segments themselves don't matter.

Segmentation analysis definitely comes into play based on the relationship you are trying to study - such as siblings for example. A cousin of mine just added a test for her brother.  They had some concerns from family stories over whether they were full-siblings or half-siblings.  A simple one-to-one comparison on Gedmatch shows that they share 2,550 cM - with nearly 1,700 cM of half-identical segments and nearly 850 cM of fully-identical segments.  This deeper look into their segments provides DNA confirmation (i.e., full siblings share 50% half-identical segments and 25% fully-identical segments) that they are full-siblings.

Ray, 23andme predicts that my Brother is a "Brother".  Just like it predicts my 3rd cousin 1x removed is "a 3rd cousin in the range of 3rd to 4th cousin".

What reason is there for excluding genealogists who are not technically minded, by requiring gedmatch, or technically minded genealogists who can not convince a DNA Tester to upload to Gedmatch from marking the profiles that clearly "further confirm with DNA a parent"?

Are you saying that if your cousin was not able to upload the raw data to gedmatch, they should not rely on a DNA Services prediction of a sibling.?

A prediction of "Brother" indicates FULL Brother.  1/2 siblings would get a different prediction.

+7 votes

I'm pretty sure my little G2G debacle last weekend is what fueled this, so I'll speak up. My recently discovered cousin and I, are on paper, 4th cousins twice removed. FTDNA placed us at a solid 3rd cousin range, with over 100cM shared. (You can see the actual difference between the GEDMatch comparison and FTDNA comparison on the other post). GEDMatch has us at about 3.8 to MRCA. But here's where it gets tricky...our common ancestral couple. Brothers married sisters. So, there's likely some pedigree collapse going on there. We show at a closer range because of that. 

It doesn't really matter to either of us, we know we're related!! But using this (along with other new matches over the last few days) is more than enough confirmation for me, that my mothers bio father is correct. That was the real issue, on my part. We have the paper trail of documents to show lineage, then the DNA only served as confirmation. 

PS. I'm desperately trying to get the other 2 new cousins to get a GEDmatch to compare the 4 of us. 

by Summer Orman G2G6 Mach 9 (94.2k points)

Here is why you should be comfortable defending your claim that your mother's bio father is correct using your DNA match to Jeanne.

Let's suppose for a moment that you and Jeanne were adopted and a DNA Service Predicts you are 4th cousins in the Range of 4th to 5 cousins. This means that guessing which side of the family relates you to Jeanne gives you about a 50% chance of being correct. If you have undisputed documentation proving you are 4th cousins 2x removed you would have been ~100% confident of the relationship, but the availability of DNA Services reduces that level of confidence. This reduction in confidence is because the documented parent may not be the biological parent.  A Non-Parental Event may have occurred. If we accept that about 5% of these well-documented profiles do not name the biological parent, then our confidence level is reduced to about ~95%.  I am using these percentages to give some context.

A single match, in most cases, should bring your confidence level back up to ~100% for those profiles up to but not including the two direct descendants of your the Most Recent Common Ancestral Couple.  The reason we exclude these two profiles from the higher confidence level is that predictions more distant than 2nd cousins are not precise enough to conclude that a Non-Parental Event does not exist for these two profiles.

You do not need to go to the segment level, to identify those profiles that are up to but not including the Most Recent Common Ancestors to be near 100% confident.

Summer, I should have connected this to the current discussion.

Your situation is a little different, but for most people who are looking to further support a 4th cousin 2x removed relationship, would do this using two additional 5th cousin matches. 1 match on the male MRCA and 1 match on the female MRCA. 

When you have a well-documented tree, then DNA is used to defend your existing claims, based on Traditional Genealogy, against contradictory claims.  

But when you don't have a well-documented tree, then you use DNA to assert a new claim.  Using only DNA matches in a well-documented tree should be sufficient to defend against virtually all other claims to the contrary.

For all Matches that do not have documentary evidence to support a specific relationship, they do need to use DNA Segments to assert a general relationship claim of sharing specific common ancestors. But this is outside the scope of the further Confirm with DNA choice found on every profile.

You are defending existing claims on some profiles, but not all. You are not asserting any claim at this time, regarding your match.

+3 votes

I have real problems following your argument.  Are you talking about your father specifically, or the father of any person, in any generation? How far back are you talking about?  (And for generations beyond your father, or back to, say, 2nd cousins, you haven't discussed the quality of the paper trail, how well it meets Genealogical Proof Standards.)

Also, the full sentence, for which you provided a partial quote, is:  "One of the many problems with utilizing small segments is that, in general, people tend to see evidence that supports their theories and reject evidence that does not."  I'm not sure how this is relevant to your argument.

by Gayel Knott G2G6 Mach 3 (33.8k points)

Gayle,  On the edit screen of every profile, there are 4 choices for the father and the same 4 choices for the mother.

  1. father/mother is non-biological
  2. father/mother is uncertain
  3. father/mother is confident
  4. father/mother is confirmed with DNA

The documentation requirements are no different from the current documentation requirements, which is why I have not discussed them.

The scenario is this…

23andme predicts that a known 3rd cousin 1x removed is a 3rd cousin in the range of 3rd to 4th cousin. 

Here is an example using Relationship Finder. http://www.wikitree.com/index.php?title=Special:Relationship&action=calculate&person1_name=Davis-30112&person2_name=Sargent-1191

Using the 23andme prediction and the documentation, I should be able to “Confirm with DNA”

  1. on my profile [Sargent-1191] confirm my father [Sargent-1192],
  2. then on profile [Sargent-1192] confirm his father [Sargent-1193]
  3. then on profile [Sargent-1193] confirm his mother [Wright-9765]
  4. then on profile [Wright-9765] confirm her father [Wright-9775]
  5. on my cousin’s profile [Davis-30112] confirm his mother [Cunningham-4523]
  6. then on profile [Cunningham-4523] confirm her mother [Wright-15973]
  7. then on profile [Wright-15973] confirm her father [Wright-9778]

but Wikitree rules don’t allow for this because it requires you to download the RAW Data so you can provide evidence to support what the prediction services have already done.

The interesting thing is that the existing rules don’t even do this. They were never intended to prove a relationship, they were only intended to give adoptees and other's looking for a connection a place to work from.

There is absolutely no reason to require every genealogist to try and recreate what the DNA Services have already provided.  There is no reason to debate whether or not small segments can be used to determine a relationship between DNA testers because of the existing requirements.  

Why does Wikitree believe they can do a better job than the DNA Services?

Ken, Where does it say you have to download your raw data to support the relationship to your father?  Or anyone else, for that matter.  My understanding is that you don't even have to do triangulation for 2nd or 3rd cousins (I'm a little squeamish, myself, about the 3rd cousins), as long as you have a good paper trail.  Can you show me in the instructions where I'm wrong?

As far as trusting the predicted relationships provided by the testing companies, I have to say, nope, no way.  My current project is trying to unsnarl several (well-triangulated, mostly well documented) matches that are predicted to be 2nd-3rd cousins, based on the total cM shared.  Paper trail says we are 5th-7th cousins.   Obviously something else going on here, but where?  And what kinds of errors would I potentially be creating by simply accepting a computer calculated relationship?

Ultimately, only people (not computers), working with multiple sources of information and also working within accepted standards, can really say what the relationships are.

Gayel, you are not wrong about  the 3rd cousin or closer DNA matches. This was a change from the original guidelines.  The 3rd cousin seems like an arbitrary decision because DNA can't really distinguish between full third cousin and 1/2 3rd cousin.  

But Genetic Genealogists do not need to know the exact relationship to map the genomes of those connected up to but not related to the Most Recent Common Ancestor(s). Not everyone has close enough relationships for this.

In your case, since your prediction is not consistent with your documentation, this would no apply to you.

I didn't actually answer your question directly about where it says you need to download raw data to support the relationship to your father.

Remember, if the closest relative you have available is a 3rd cousin 1x removed, you are forced to download your raw data because of the wikitree requirements that you either match on the x chromosome, or the necessity to identify a triangulated segment using an additional match.

The question Wikitree asks on each profile is "Does the resulting available information about a match further support the existing evidence that this person is the father or the mother?

I have no doubt that a prediction by the DNA Service that is consistent with the existing evidence can be used as genealogical evidence. When the result is not consistent, then additional testing is needed to explain the result and can not be used to support the claim unless the result can be explained.

International Society of Genetic Genealogy - Autosomal DNA

Accuracy of tests

"Autosomal DNA tests can be used to confirm relationships with a high level of accuracy for parent/child relationships and all relationships up to the second cousin level. For all relationships other than parent/child relationships additional contextual and genealogical information is required to confirm the nature of the relationship.

"For genealogical relationships between second cousins once removed and 5th cousins a more careful approach is necessary and data needs to be collected from multiple family members." 

Multiple family members = triangulation.

23andme and FTDNA provide a user option to "Confirm" the relationship between the DNA tester and a Match.  This is NOT what the user functionality provided by Wikitree.  If Wikitree wanted a way to "Confirm" this relationship, it would not have put the question on EVERY profile. It would have been placed on a similar screen as Relationship Finder with a single checkbox "Confirmed with DNA", and the exact methods of proof with the exact templates that are argued for today. ALL the references to articles relate to this functionality. We are not being asked to confirm relationships between DNA testers.

If this was the Wikitree objective, then NO ARGUMENT from me. BUT this is NOT what we are being asked!

A single match on a documented tree gives us significant information that does not change with the additional testing currently required.  In most cases, the profiles up to but not including the Most Recent Common Ancestors can have their segments mapped. If they can be mapped, then there is virtually no possibility of a NonParental Event. This is worth noting by "Confirming with DNA" on just those profiles.

The additional testing that requires triangulation is directed solely at the last two profiles. This is what would be necessary to confirm a relationship between DNA testers but I repeat, this is not what we are being asked.

The important thing to note is that the additional testing I am told is required has absolutely no effect on the results of the single Match.
+3 votes
Ken,

I understand where you're coming from and it's indeed true that there are two types of methods how atDNA is used at WikiTree (as you described).

However you're simplifying things too much and that is a dangerous thing that can lead to many wrong conclusions. The randomness of DNA and endogamy make it very difficult to predict relationships beyond even 2nd cousins and there are many examples in the recent days being posted on DNA genealogy mailing list where with the recent change of algorithm (to the claimed more precise) by AncestryDNA moved 2nd cousins (by paper trail) to 3rd or 4th cousins and vice-versa.

Yes, everyone would like to have a good relationship predictor but so far no one has (hence the 3 testing companies are giving ranges, like 4th cousin and beyond).

It's absolutely ok to follow your example and verify a parent/child or grandparent/child or aunt/uncle/nephew/niece relationship with the amount of shared DNA. Even the half variants in there.

But as soon as we hit 1st cousins (estimated 12.5% DNA shared) we're in troubled waters, based on 662 reported 1st cousins relationships Blaine Bettinger has reported the average cM to be 881 cM (850 cM would be 12.5%) but the range reported is from 83 cM (which is where 2C2R, 3C1R and even 3C are reported on average) to 1559 cM (obviously through an endogamous family) which is on the average of a aunt/uncle/grandchild/niece/nephew relationship.

You see the problem that your method creates?

As you rightfully pointed out, people are quick to make assumptions that fit their theory. Like you wrote in another G2G answer, a Y-DNA test doesn't distinguish between your assumed father or one of his brothers to be your genetic father (as they both have the same Y-DNA - at least to what we can analyze with current technology).

I think we all agree that there is a big challenge in those that have endogamy in their ancestry (like Jewish, German, Russian, Italian and other nationalities that immigrated into the US) to sort out relationships. Dr. Tim Janzen who is one of the well-known experts on endogamy and DNA has many examples where the relationship predictor of those companies that invested millions were completely wrong.

Now you're saying we should take this to prove family trees?
by Andreas West G2G6 Mach 7 (74.8k points)

Andreas, Thank your response.

I believe we don't agree on the actual objective and these seem to be at the opposite end of the spectrum.

From what I read on wikitree, we are to make a choice that indicates that there is DNA evidence to support a claim that a particular person is the father or mother of another particular person. Just like any other genealogical evidence, it supports but does not conclude.

My example of a distant male relative working for over a year on my house would share the same ySTR as me, my brother and a known 7th cousin.  I understand that my son could be my nephew, 1st cousin, 7th cousin, something in between or no blood relation at all. We are making conclusion, we are collecting evidence.

I do accept though that a ySTR match supports the documented relationship between myself, and my father, as well as all those, connected to my match but it does not represent a conclusion. Other evidence may provide sufficient weight to refute this claim.  auDNA is not much different.

With this in mind, when a DNA service prediction is consistent with the existing documented tree, this can be used as just a piece of evidence, like any other geological evidence, to support a singular claim of some parentage.  My thoughts are that if you identify a profile that could map a segment to a particular profile, then this is a single piece of evidence. 

I have never claimed that all families will be able to take advantage of this and will have to look deeper, but so far, my family tree and the predictions have been accurate. I am not claiming that a prediction that is not consistent with the documented tree should be used to refute a claim. If the prediction and the documentation do not agree with each other, then you can't use it to support a claim.

Your objective appears to me to be to prove an exact relationship between two DNA testers.  Even a ySTR match only provides support for a half cousin relationship, but not a full relationship.  A single auDNA match for beyond 2nd cousin relationships, auDNA only supports 1/2 cousin relationships. But in these cases, we know that segments for all those profiles up to but not including the Most Recent Common Ancestor could be mapped.  These maps are both useful and can be used to support a claim, even if they are not conclusive.

Also, nothing in Wikitree states or even suggests that you are to provided evidence of the relationship of a match. 

If you and your match are both men there is a special opportunity for one-to-one confirmation if you match either on your X chromosomes. Since men have only one X chromosome, no triangulation is necessary.” see help

It doesn’t make any difference if the prediction indicates no match, or that no auDNA segments of any size exist. No triangulation is necessary.

Why does anyone care about segments smaller than 7 cM.

“Find segments measuring 7 cM or more that you share (i.e. overlap) with two or more distant cousins.

Those who all match each other on the same segment form a triangulated group (TG). Using WikiTree's Relationship Finder find the shared ancestral couple for that TG.” see help

According to this, it doesn’t matter what the relationship is, as long as there is at least one 7cM triangulated segment, you can find the shared “Ancestral Couple”. This all that is needed to confirm all the profiles.

Related questions

+58 votes
4 answers
2.1k views asked Mar 14, 2016 in The Tree House by Kitty Smith G2G6 Pilot (638k points)
+18 votes
1 answer
+17 votes
3 answers
876 views asked Dec 3, 2015 in The Tree House by Peter Roberts G2G6 Pilot (695k points)
+13 votes
3 answers
+7 votes
1 answer
257 views asked Dec 12, 2017 in The Tree House by Kerry Larson G2G6 Pilot (233k points)
+27 votes
2 answers
+12 votes
2 answers
203 views asked Jul 18, 2017 in The Tree House by Peter Roberts G2G6 Pilot (695k points)
+18 votes
2 answers
+10 votes
0 answers

WikiTree  ~  About  ~  Help Help  ~  Search Person Search  ~  Surname:

disclaimer - terms - copyright

...