> From: Linda Boorom > Sent: Saturday, October 05, 2013 11:59 AM > I'm thinking I need to go back a generation or 2 & find some more distant > cousins to test? I'm a newbie at genetic genealogy but it seems to me the goal (assuming limited resources) would be to test ... 1) yourself 2) one of your parents Then any time *you* hit an atDNA match with someone, you immediately know which side of the family shares a joint ancestor with your match, depending on whether the match also hits (2). So the next goal would be to be able to figure out which of your four grandparents' ancestors' lines the match is in. To do that, you need to also have on hand another test from one of those two lines, and the binary answer to "does your match hit that person too" answers your question. So you also need to test ... 3) your other parent 4) a cousin of one of your parents (your 1st cousin 1x removed), or oldest available descendant thereof 5) a cousin of your other parent, or oldest available descendant thereof So now you can determine which of your four grandparents' ancestries your match is descended from. And next, you'd want to be able to determine which of your eight great-grandparents your match is in, which would require testing ... 6) the oldest available descendant of any of your maternal grandmother's cousins (your 1st cousin 2x removed) 7) the oldest available descendant of any of your maternal grandfather's cousins 8) the oldest available descendant of any of your paternal grandmother's cousins 9) the oldest available descendant of any of your paternal grandfather's cousins And so on; i.e. to the extent that you have information about who these people are, your next 8 steps should be to test one descendant of each of your 8 great-grandparents' cousins ... All other things being equal, wouldn't this strategy be the fastest/highest-yield in terms of being able to determine which of your x-great grandparents is the point where your ancestry intersects your match's? I.e. better this approach than trying to test any more than just one person among a line's descendants? (I.e. assuming you have tested *at least* one of your parents, you'd get more value out of testing persons (4) and (5) above than out of testing any of your aunts and uncles and cousins?) I guess this strategy presumes all relationships are biological (because it assumes that if there's not a match with a descendent on one side of a binary tree, there necessarily would be a match with someone on the other side) ... and aren't generated within endogamous populations ... But so, at what point would this strategy have a statistically important chance of breaking down (because it's a very unsure thing whether the person you've managed to test has much likelihood of share any DNA with your match)? Best, Eric

Eric This process is a good one, but it won't cover all your Matches at each generation. At the parent generation, your Matches fall into 4 categories: Parent 1, Parent 2 either Parent 1 or 2 (when the parents are related) and IBS (not from either parent) - these latter two you recognized in your post. Aunts/uncles will help some here, if you can't test a parent. At the grandparent generation, a first cousin only shares 25 percent of the atDNA of the grandparent, and so will only match a fraction of the Matches from that grandparent with other cousins (prospective Matches). And due to randomness, even 8 widely separated first cousins wouldn't provide full coverage. A second cousin only has 12.5 percent of each great grandparent's atDNA and the coverage drops (by a factor of 4), with each succeeding generation. And this presumes you are only interested in mapping your own atDNA. Because you got only half of each parent's atDNA, you will not match with roughly half their Matches. However, your process will provide a lot of help, and give you good coverage over all your ancestry. I'll add that, although the plan is good, it is not as easy as it appears. The break down occurs in the genealogy arena. Most of your Matches will be greater than 5th cousins and few people have good genealogies that cover that area. Look at your own genealogy to see what percent of 6G grandparents you know (they are need for 7th cousins). If you and your Match each know 50 percent of those ancestors, the two of you will find a Common Ancestor 1/4 of the time. So to your point - your process will help map the chromosomes at the first few generations (and this is very helpful), but determining the Common Ancestor for most Matches will still be a challenge. Jim - Sent from my iPhone - FaceTime! On Oct 8, 2013, at 5:29 AM, "Eric S Johnson" <crates@oneotaslopes.org> wrote: >> From: Linda Boorom >> Sent: Saturday, October 05, 2013 11:59 AM > >> I'm thinking I need to go back a generation or 2 & find some more distant >> cousins to test? > > I'm a newbie at genetic genealogy but it seems to me the goal (assuming > limited resources) would be to test ... > 1) yourself > 2) one of your parents > Then any time *you* hit an atDNA match with someone, you immediately know > which side of the family shares a joint ancestor with your match, depending > on whether the match also hits (2). So the next goal would be to be able to > figure out which of your four grandparents' ancestors' lines the match is > in. To do that, you need to also have on hand another test from one of those > two lines, and the binary answer to "does your match hit that person too" > answers your question. So you also need to test ... > 3) your other parent > 4) a cousin of one of your parents (your 1st cousin 1x removed), or > oldest available descendant thereof > 5) a cousin of your other parent, or oldest available descendant > thereof > So now you can determine which of your four grandparents' ancestries your > match is descended from. And next, you'd want to be able to determine which > of your eight great-grandparents your match is in, which would require > testing ... > 6) the oldest available descendant of any of your maternal > grandmother's cousins (your 1st cousin 2x removed) > 7) the oldest available descendant of any of your maternal > grandfather's cousins > 8) the oldest available descendant of any of your paternal > grandmother's cousins > 9) the oldest available descendant of any of your paternal > grandfather's cousins > And so on; i.e. to the extent that you have information about who these > people are, your next 8 steps should be to test one descendant of each of > your 8 great-grandparents' cousins ... > > All other things being equal, wouldn't this strategy be the > fastest/highest-yield in terms of being able to determine which of your > x-great grandparents is the point where your ancestry intersects your > match's? I.e. better this approach than trying to test any more than just > one person among a line's descendants? (I.e. assuming you have tested *at > least* one of your parents, you'd get more value out of testing persons (4) > and (5) above than out of testing any of your aunts and uncles and cousins?) > > I guess this strategy presumes all relationships are biological (because it > assumes that if there's not a match with a descendent on one side of a > binary tree, there necessarily would be a match with someone on the other > side) ... and aren't generated within endogamous populations ... > But so, at what point would this strategy have a statistically > important chance of breaking down (because it's a very unsure thing whether > the person you've managed to test has much likelihood of share any DNA with > your match)? > > Best, > Eric

Jim, Elizabeth--I hear you (and understand why: 50%, 25%, etc.) about how the strategy I've proposed will have potential holes in it. But the original question (in which I'm very interested, too) was about the *optimal* strategy to choose "who to test next," not "how to have a set of tests which are collectively perfect." If I've tested myself, and my two parents (and none of my grandparents are alive), and I only have enough resources to test one more person, wouldn't it be smartest to spend that a parent's cousin ... etc. (up through the chain I described). If I have lots of resources, then the time will come when my strategy's exhausted itself (because I won't have enough information to find person number 11), and *at that point* the highest marginal value for "the next test" would come from another cousin of one of my parents, etc. ... ? I'm not at all sure of myself--just trying to apply my own primitive logic to the question. Best, Eric OpenPGP: 0x1AF7E6F2 ● Skype: oneota ● XMPP/OTR: berekum@jabber.ccc.de ● Silent Circle: +1 312 614-0159 > -----Original Message----- > From: autosomal-dna-bounces@rootsweb.com [mailto:autosomal-dna- > bounces@rootsweb.com] On Behalf Of Jim Bartlett > Sent: Tuesday, October 8, 2013 21.57 > To: autosomal-dna@rootsweb.com > Subject: Re: [AUTOSOMAL-DNA] Matches with a common surname > > Eric > > This process is a good one, but it won't cover all your Matches at each > generation. > > At the parent generation, your Matches fall into 4 categories: Parent 1, > Parent 2 either Parent 1 or 2 (when the parents are related) and IBS (not > from either parent) - these latter two you recognized in your post. > Aunts/uncles will help some here, if you can't test a parent. > > At the grandparent generation, a first cousin only shares 25 percent of the > atDNA of the grandparent, and so will only match a fraction of the Matches > from that grandparent with other cousins (prospective Matches). And due to > randomness, even 8 widely separated first cousins wouldn't provide full > coverage. > > A second cousin only has 12.5 percent of each great grandparent's atDNA > and the coverage drops (by a factor of 4), with each succeeding generation. > > And this presumes you are only interested in mapping your own atDNA. > Because you got only half of each parent's atDNA, you will not match with > roughly half their Matches. > > However, your process will provide a lot of help, and give you good coverage > over all your ancestry. > > I'll add that, although the plan is good, it is not as easy as it appears. The > break down occurs in the genealogy arena. Most of your Matches will be > greater than 5th cousins and few people have good genealogies that cover > that area. Look at your own genealogy to see what percent of 6G > grandparents you know (they are need for 7th cousins). If you and your > Match each know 50 percent of those ancestors, the two of you will find a > Common Ancestor 1/4 of the time. > > So to your point - your process will help map the chromosomes at the first > few generations (and this is very helpful), but determining the Common > Ancestor for most Matches will still be a challenge. > > Jim - Sent from my iPhone - FaceTime! > > On Oct 8, 2013, at 5:29 AM, "Eric S Johnson" <crates@oneotaslopes.org> > wrote: > > >> From: Linda Boorom > >> Sent: Saturday, October 05, 2013 11:59 AM > > > >> I'm thinking I need to go back a generation or 2 & find some more > >> distant cousins to test? > > > > I'm a newbie at genetic genealogy but it seems to me the goal > > (assuming limited resources) would be to test ... > > 1) yourself > > 2) one of your parents > > Then any time *you* hit an atDNA match with someone, you immediately > > know which side of the family shares a joint ancestor with your match, > > depending on whether the match also hits (2). So the next goal would > > be to be able to figure out which of your four grandparents' > > ancestors' lines the match is in. To do that, you need to also have on > > hand another test from one of those two lines, and the binary answer to > "does your match hit that person too" > > answers your question. So you also need to test ... > > 3) your other parent > > 4) a cousin of one of your parents (your 1st cousin 1x removed), or > > oldest available descendant thereof > > 5) a cousin of your other parent, or oldest available descendant > > thereof So now you can determine which of your four grandparents' > > ancestries your match is descended from. And next, you'd want to be > > able to determine which of your eight great-grandparents your match is > > in, which would require testing ... > > 6) the oldest available descendant of any of your maternal > > grandmother's cousins (your 1st cousin 2x removed) > > 7) the oldest available descendant of any of your maternal > > grandfather's cousins > > 8) the oldest available descendant of any of your paternal > > grandmother's cousins > > 9) the oldest available descendant of any of your paternal > > grandfather's cousins And so on; i.e. to the extent that you have > > information about who these people are, your next 8 steps should be to > > test one descendant of each of your 8 great-grandparents' cousins ... > > > > All other things being equal, wouldn't this strategy be the > > fastest/highest-yield in terms of being able to determine which of > > your x-great grandparents is the point where your ancestry intersects > > your match's? I.e. better this approach than trying to test any more > > than just one person among a line's descendants? (I.e. assuming you > > have tested *at > > least* one of your parents, you'd get more value out of testing > > persons (4) and (5) above than out of testing any of your aunts and > > uncles and cousins?) > > > > I guess this strategy presumes all relationships are biological > > (because it assumes that if there's not a match with a descendent on > > one side of a binary tree, there necessarily would be a match with > > someone on the other > > side) ... and aren't generated within endogamous populations ... > > But so, at what point would this strategy have a statistically > > important chance of breaking down (because it's a very unsure thing > > whether the person you've managed to test has much likelihood of share > > any DNA with your match)? > > > > Best, > > Eric > > > > ______________________________ > For answers to Frequently Asked Questions about mailing lists, please see: > http://dgmweb.net/MailingListFAQs.html > > > ------------------------------- > To unsubscribe from the list, please send an email to AUTOSOMAL-DNA- > request@rootsweb.com with the word 'unsubscribe' without the quotes in > the subject and the body of the message

This is a logical strategy, but based on my own experience I think you'll get more information by testing your parents, if living, and as many of their first and second cousins as you can find and persuade. Let the more distant relatives accumulate from matches to other people you find through the testing company. I've tested seven of my mother's paternal first cousins, plus one person who is their mutual second cousin. In aggregate the first cousins' results are more informative than seven more distant relatives would be, because they also match each other in various combinations, permitting deductions about who got what from each of their grandparents. For example, if my mother matches first cousins A and B, but not second cousin Z, we can't conclude much from that information alone. But if we also know that first cousins C and D match each other over the same region of the same chromosome, and do match second cousin Z (who is related to their grandfather), then we also know that my mother, A and B inherited that chromosomal segment from their grandmother. This analysis works best in my experience if you use the raw data to determine the exact endpoints of matching segments. That's somewhat tedious, but still quite feasible. Elizabeth Harris ncgen@mindspring.com On Oct 8, 2013, at 5:29 AM, Eric S Johnson wrote: >> From: Linda Boorom >> Sent: Saturday, October 05, 2013 11:59 AM > >> I'm thinking I need to go back a generation or 2 & find some more distant >> cousins to test? > > I'm a newbie at genetic genealogy but it seems to me the goal (assuming > limited resources) would be to test ... > 1) yourself > 2) one of your parents > Then any time *you* hit an atDNA match with someone, you immediately know > which side of the family shares a joint ancestor with your match, depending > on whether the match also hits (2). So the next goal would be to be able to > figure out which of your four grandparents' ancestors' lines the match is > in. To do that, you need to also have on hand another test from one of those > two lines, and the binary answer to "does your match hit that person too" > answers your question. So you also need to test ... > 3) your other parent > 4) a cousin of one of your parents (your 1st cousin 1x removed), or > oldest available descendant thereof > 5) a cousin of your other parent, or oldest available descendant > thereof > So now you can determine which of your four grandparents' ancestries your > match is descended from. And next, you'd want to be able to determine which > of your eight great-grandparents your match is in, which would require > testing ... > 6) the oldest available descendant of any of your maternal > grandmother's cousins (your 1st cousin 2x removed) > 7) the oldest available descendant of any of your maternal > grandfather's cousins > 8) the oldest available descendant of any of your paternal > grandmother's cousins > 9) the oldest available descendant of any of your paternal > grandfather's cousins > And so on; i.e. to the extent that you have information about who these > people are, your next 8 steps should be to test one descendant of each of > your 8 great-grandparents' cousins ... > > All other things being equal, wouldn't this strategy be the > fastest/highest-yield in terms of being able to determine which of your > x-great grandparents is the point where your ancestry intersects your > match's? I.e. better this approach than trying to test any more than just > one person among a line's descendants? (I.e. assuming you have tested *at > least* one of your parents, you'd get more value out of testing persons (4) > and (5) above than out of testing any of your aunts and uncles and cousins?) > > I guess this strategy presumes all relationships are biological (because it > assumes that if there's not a match with a descendent on one side of a > binary tree, there necessarily would be a match with someone on the other > side) ... and aren't generated within endogamous populations ... > But so, at what point would this strategy have a statistically > important chance of breaking down (because it's a very unsure thing whether > the person you've managed to test has much likelihood of share any DNA with > your match)? > > Best, > Eric > >