Sunday, September 24, 2017

Our Genetic Portrait - Part Two

In my post last week, I introduced DNA Painter and described how I identified sixty segments that my family received from specific ancestors of my father. The ancestors concerned are three of my father's grandparents, two of his great-grandmothers and one second great-grandfather. The segments are pretty much low-hanging fruit - easy to identify and confirm.

Reminder - this profile is called "Israel Pickholtz family," not just "Israel Pickholtz." I saw no reason to use only my own segments when my brother and four sisters have segments from those same ancestors which can only enhance my map. That decision required calling this profile "female" because my sisters have a paternal X, even though I myself do not.

Mapping my mother's side
My mother's side is quite a different story. Not only do we have few known ancestral lines, but we have no test results from third or fourth cousins. Since all we have are second cousins (plus two first cousins, Kay and Leonard), I can only use DNA Painter to attribute segments to my grandparents, Rachmiel Gordon and Sarah Rosenbloom. I have no idea what they received from which of their parents.

There are now 132 identified segments on my mother's sides, which is an update since last week.

On the Gordon side, we have two second cousins on GEDmatch - Judy and Ruth. They are first cousins and their grandmother is Rachmiel Gordon's older sister. Andrea, the daughter of Judy and Ruth's first cousin, has also tested is new to GEDmatch and she adds nine segments that Ruth and Judy do not have.. And there are two third cousins who tested but have declined to upload to GEDmatch.

We have six tests on the Rosenbloom side, three from my grandmother's older sister Alta and three from her younger brother Hymen (Chaim Benzion). Here is the  structure of the family - both sides together. Those who have tested and uploaded to GEDmatch are in yellow.

The Gordons
I started with Judy and Ruth. Judy has twelve matches with us - "us" being the eight grandchildren of my grandparents. Ten of those twelve have enough good matches that I consider them certain. On chromosome 5, Judy has a match with Kay and a similar-sized match with my sister Sarajoy. Normally I would reject that as being not good enough, but the fact that Kay's father has no Jewish DNA gives her a bonus since there is no endogamy from that side. Nonetheless, I rated this match "very likely" rather than "certain."

On chromosome 9, Judy has a match with me, a similar match with my sister Judith and a much smaller match with my brother Dan. I would like to see more people here, especially either Kay or Leonard, so here too I rated it "very likely."

Next I looked at Ruth's matches with us, ignoring those which she shares with Judy and which I had already recorded. There are fifteen. Ruth's total match with me is nearly 500 cM, much higher than we would expect from a second cousin, and I suspect that her father has significant Galitzianer in his background. (His parents died when he was young, so we really don't know.) Here Kay and Leonard become more important. Matches which include them, I regard as certain . Matches which do not, I downgraded to "very likely" because there is a non-negligible possibility that this such matches come from our fathers rather than our Gordon mothers.

I also downgraded a match between me and Ruth because although Leonard is there, his match is much smaller. It simply doesn't feel certain. All told, Ruth has eight certain matches with us and seven "very likely."

Andrea has nine matches but six of them are only with my mother's children. I marked those as "very likely" rather than certain. I left a comment in those cases.

This is the set of matches between Ruth, Judy and Andrea and my grandparents' eight tested grandchildren.

The Rosenblooms
A bit of background on my grandmother Sarah's family. She and her sisters and brother were born in Borisov in what is now Belarus. Her father Israel David Rosenbloom appears in the 1874 revision list in Borisov so I assume that's where his family had lived. His wife Etta Bryna - whose surname is unknown - could have been from anywhere.

My grandfather Rachmiel Gordon was born in Dolginov which is not far away and it is possible, even likely, he had some family in Borisov. Sarah's sister Alta married Berl Kaplan whom we think was from Borisov, but we are not sure. Uncle Hymen married in Pittsburgh - to a woman from Schedrin, also Belarus, but over a hundred miles from Borisov.

So I decided that I could not set matches that did not include any of Uncle Hymen's grandchildren as "certain" due to the possibility that there was some kind of connection between Sarah and Alta's husbands' families.

But it is more complicated yet. Louis Jaffe, Judy and Ruth's grandfather, was also from Borisov, so I had to be on the lookout for matches that might have included his family mixed in with the Rosenblooms.

I will spare you all the detail, but there are a lot of segments that required analysis and decisions. These are the results of the six Rosenbloom second cousins - three from Alta's granddaughters and three from Uncle Hymen's grandchildren. (Hovering over and clicking on the segments gives much additonal information.)

Lots of segments here. Quite a few overlapping segments as well, because a match with one of Alta's granddaughters may be similar to but different from a match with one of Uncle Hymen's grandchildren. This is not trivial, even if there is some (unnecessary) duplication.

If you look at the segment at the far right end of chromosome 19, you will find a similar match on the Gordon set above. Here is the chromosomal view.

The green match is from Judy. The brown match is from Lydia and Inna.

These are not distinct Gordon and Rosenbloom matches. Judy triangulates here with Lydia and Inna. Like some kind of general Borisov segment. And who know how many more are similar, but we cannot identify them. I rated it as "very likely" but I'm not even sure what that means in this context.

And can you imagine, there are people who would do this automatically!! Endogamous populations are hard to do well. Anyone can do them badly.

Monday, September 18, 2017

Our Genetic Portrait - Part One

DNA Painter
A few months ago, I came to the conclusion that I have been spending entirely too much time on the trees and not enough on the forest. I would hear from people who think they may be related to my family and I'd look at their GEDmatch kits, comparing them to my 110+ project members looking for segments with multiple matches of 10 centiMorgans or more to see what patterns would show up.

Most of them showed some direction, but nothing really specific, appearing mostly to be too far back in time to do anything with, especially without the benefit of common surnames, geography or multiple family kits

I decided I needed to do some chromosome mapping in order to attribute particular chromosomes to specific ancestors. I had used Kitty Cooper's wonderful Chromosome Mapper a few years ago, but I was far from comfortable with it. I circled the Chromosome Mapping course at GRIP next June and wondered if it would relate to the specific issues of endogamy.

A couple of weeks ago, someone on Facebook mentioned a new tool called DNA Painter and I decided to give it a try. This is a report on my first efforts. The chart below shows my first 183 segments - 60 on my father's side and 123 on my mother's side. And there will be more.

In the course of creating this map, I have been in contact with the developer and he has welcomed my comments and questions. There is a lot to like here.

The data comes from the GEDmatch one-to-one and X-one-to one results.

What you are probably supposed to do is take all your matches with a particular person and copy them into the Painter, but this is not a responsible way to deal with families from endogamous populations.

For us, we have to go segment by segment, analyzing each one, looking for matches with multiple people that are based on more than "I match cousin so-and-so, so we must be related IN THE MOST OBVIOUS WAY." Not necessarily. I'll get into that in greater detail on my mother's side.

First note that this profile is called "Israel Pickholtz family," not just "Israel Pickholtz." I saw no reason to use only my own segments when my brother and four sisters have segments from those same ancestors which can only enhance my map. That decision required calling this profile "female" because my sisters have a paternal X, even though I myself do not.

I decided to take the next, obvious step - the ancestral matches of my father's sister and brother, Aunt Betty and Uncle Bob, since all their ancestors are also my ancestors. But their matches are not the same as my father's matches (as represented by his children), so there will be places where the final map will include segments from both my grandfather and my grandmother AT THE SAME PLACE. This is a good thing, because it fills out the ancestors, but it will require extra care when some new person comes into the picture with a match that needs defining.

Mapping my father's side
I began with my fourth cousin Anna, the great-granddaughter of Uncle Selig about whom I have written several times.
This chart may be familiar to you from my DNA presentations.
The circled matches on Anna's chromosomes 3, 8 and 15 are definitely from our second great-grandmother Rivka Feige Pikholz and are the basis of the demonstration that Anna and her half-brother David are descendants of Uncle Selig. I also included Anna's matches on chromosome 21, as these are clearly from the same source.

But I tend to be conservatiive, so on chromosome 15 where Marty and Anna share 50 cM, I decided to record as certain only the 22.6 cM that Anna shares with Marty and Herb. The rest of Anna and Marty's long segment I recorded as "possible," though I could have chosen "very likely." Here is how that segment appears in full.

Another set of Rivka Feige matches came from my father's half second cousin Lia who contributed five segments to the map. These are the Rivka Feige segments from Anna and Lia.

Note that on chromosome 15, one segment is pink rather than red. That is Marty's "possible" segment.

The second easy match was with Debbie from North Carolina, whom I wrote about in May 2016. She has one large match with us on chromosome 2 and and another on the X. The first is definitely from my second great-grandmother Mari Zelinka and the second is either from her or from my grandmother's Bauer/Stern side, but since I have seen no other indication this might be the case, I attributed the segment to the Zelinkas and called it "very likely."

Together with two segments from my fourth cousin Milan in Prague, this is what we have from Mari Zelinka.

There is also a single segment in purple which we received from Mari Zelinka's father - based on a match with my fifth cousin Cyndi Norwitz on chromosome 18.

Two other easily identifiable ancestors on my father's side are my great-grandmothers Regina Bauer and Jute Lea Kwoczka. The first I established using fifteen matching segments with my father's second cousin Shabtai and the second using nineteen segments with my father's second cousin Bruce and another eleven with my third cousin Pinchas. Here is how those appear.

This gives me a total of sixty segments of low-hanging fruit on my father's side, supposedly representing 36% of my the ancestral DNA from that side.

But in fact that is not correct. As you can see clearly on chromosomes 3, 8 and 18 there are overlapping ancestors. Different versions of the same segment from different ancestors. Here is where I have to be very careful when I want to look at matches with new people. And the 36% is nםt representative of all the ancestors on my father's side - it just refers to segments which have at least one ancestor represented.

Perhaps I shall redo the map later, so I can show a separate set of chromosomes for each of my paternal grandparents.

My mother's side is a different story entirely and I hope to tell you about that next week.

For now I shall wish you all a happy, healthy, peaceful and prosperous 5778 and that you and your families should be written in the Book of Life.

לשנה טובה תכתבו ותחתמו

Housekeeping notes
When I visited the cemetery in Vienna a few months ago, I reported that several of the family graves had no tombstones or had tombstones in very bad condition. I was kind of hoping that close family members might address this and I can now report that one actually did. The grandson of Isidor Riss sent me photographs of the stone he put up for Isidor (who died in 1937) and his wife Ernestina who was killed in Auschwitz. 

In other news, I had a look at my Ancestry DNA matches for the first time in a long time and dropped a note to a woman named Andrea whom I didn't recognize but who seems to be fairly closely related. Turns out she is the daughter of a second cousin on my mother's paternal side. I knew her name but we have never met. I asked her to join GEDmatch and she said she would. Meantime we are now Facebook friends.

Sunday, September 17, 2017

Issues With GEDmatch - A New Example

Last night, I received an inquiry from Banai Feldstein, whom I am sure many of you know, reporting that her mother has a new match with one of my people. His name is Marty and his Family Finder results came in maybe ten days ago. He is her fourth closest match over all on GEDmatch and she thought it worth a look. 
Family Tree DNA is in the same ballpark.

Marty is my second cousin on the Pikholz side. Marty's sister Rhoda and first cousin Roz do not appear on Banai's mother's match list at all. (Banai herself does match Rhoda and Roz, which indicates something from her father.)

Banai's mother's match of 22.4 cM with Marty overlaps a match with a match of 13 cM with Herb who is a first cousin to our fathers - so there seems to be something on our joint Pikholz/Kwoczka side. But none of the other cousins are there and it appears fairly obscure.

I did a one-to-one and the results were different.
The total of 133.6 became 70.4 and "estimated generations to MRCA" went from 3.4 to 3.8. This is not a threshhold thing - or if it is, it is pretty obscure. If I drop the threshhold to 4, total cM is 141.6 and at 5, it's 124.5 cM.

So we decided to leave it for now, but this is another internal inconsistency with GEDmatch data. Their stack has become worrying and it cannot serve their image well among users.