Wednesday, February 22, 2017

23andMe Has The Best DNA Tools: Advice On How To Use Them

Chromosome map

The tools for working with matches at 23andMe are outstanding. The only drawback is the lack of response from matches. There are no trees at 23andMe, and few people have posted ancestral surnames. I'm finding, however, if I can find a common ancestor with a 3rd cousin match I can identify segments which lead to resolving how many of my matches relate to my mother and I.

I've been able to identify 3rd cousin matches by searching for trees at Ancestry.com. I search using the member directory. Most 23andMe matches don't have Ancestry trees, however. They generally don't have trees at MyHeritage either. Most testers at 23andMe have little ancestral information, if any at all.

The best way to ID 3rd cousin, or 2nd cousins for that matter, is compare segments using Genome Mate Pro, or a spreadsheet. If you compare your matches at 23andMe with GEDmatch cousins, for instance, you'll find matches who have tested at all 3 major companies. Since AncestryDNA matches often have some sort of tree you can sometimes find matches with trees who share the same segment, or overlapping segment, with a 23andMe match. You may then be able to identify these segments as coming from common ancestors.

You can compare segments with all your Family Tree DNA matches, if you've tested with them, or transferred your AncestryDNA raw data over to them. You will find more trees and surnames at Family Tree DNA than you'll find at 23andMe. You may find segment matches and common ancestors comparing here.

Once you have a map with identified segments 23andMe matches will be easier to identify, even if they aren't sharing with you, or open sharing. The relatives in common feature will allow you to identify matches. I didn't copy the names for privacy reasons in the screenshots below.

The first shot below is from the common relatives feature, at 23andMe, displaying matches shared with a match I selected. This match is a 3rd cousin. The shared matches are 3rd cousins with me; as stated below in the first column. The percentage of DNA shared with me is also given. Column two gives the relationship of these shared matches with the match I selected. The first common match is actually the mother of  the match I selected. The second common match is the uncle of the selected match, as stated below. The uncle isn't sharing with me, and has a different surname. I wouldn't have guessed their relationship with out the same surname. The fact the relationship is provided is so helpful. I have had contact with the person described as mother here. We have found common ancestors. So even without any surname information posted by these three I have determined how they are related. I have the shared segments for the mother and daughter and can marked them with the shared ancestral couples' surnames. These segments have helped me identify many more matches. I've been able to confirm some of these at Ancestry.com, with trees. I don't have the uncle's segments, but know how he is related.

 

My mother and I don't have 2nd cousin matches at 23andMe so I'm focusing on 3rd cousin matches with substantial shared amounts of DNA. I'm using the large segments to infer how those sharing smaller segments might be related. Below you see a 3rd cousin match who is a 1rst cousin to the match I selected to compare with. This would definitely lead me to believe we three share common ancestors. Both are sharing with me and share DNA in some of the same places on the chromosomes. I've marked their shared segments and again have identified many shared matches who match in the same place on the chromosomes.

Relatives in common feature at 23andMe
 
 
Below is a screenshot showing a 3rd cousin's stats who isn't open sharing, but is a common match with another match I selected. You can see I've sent them a request to share. Oddly this match shares a higher percentage of DNA, but is described as the selected matches' 4th cousin. The high percentage of DNA that we all share would lead me to believe there is a good likelihood that we are all related through the same ancestors. Below this good match I have a shared 4th cousin match who is described as a distant cousin to the selected match. They share very little DNA and the likelihood we all share the same common ancestors would be less likely. 


 
In the example below the match I selected is an open sharing match with everyone, as is one of our shared matched.  Below you can see  the stats for our common open sharing match. The "yes" next to this match means we share a DNA segment in the same place, and we can compare in the chromosome browser. This match has no posted information about their ancestors, or surnames.  I have identified common ancestors with the selected match so I can infer this common match shares these same ancestors. 23andMe tells us these shared matches do indeed match each.
 
Here are the segments they share with me. They overlap and they match each other in the same place.



I have other matches in the same place on chromosome 10 who have the same shared ancestors. I can safely name the region after the shared Roller/Zirkle ancestors, who are the most common recent ancestors, MRCA. When I have matches with segments on chr 10, on my maternal side, I can assume if this is a true match, they must be related through the Roller line.
 
I have been able to nearly completely fill in chromosome 14 for my mother with segments shared by tree verified matches on her paternal side (her mother was Nicaraguan, and her father Anglo American). I don't share these matches on chr 14 with my mother. Looking at my own segment map I have Nicaragua matches here, not the Anglo American matches my mother has. I didn't inherit DNA from my Grandfather on ch 14. I cannot inherit segments from both of my paternal, or maternal, grandparents in the exact same place on a chromosome.
 
 
 
The 23andMe tools for atDNA are a joy to work with. They greatly aid us in achieving our goal of identifying how we are related to someone in an accurate way. You can't accurately identify how distant cousins are related without a segment map. We can be related to distant cousins through more than one line. A good segment map can find errors in our inferred DNA relationships, and point us to the correct DNA connection. The fact 23andMe shares predicted relationships between common matches also helps us solve relationship puzzles. Many have tested close relatives, we can identify these family group matches when we see their relationship to each other in the common relative lists. They may not share a common surname, but may be parent and child , or uncle and niece or nephew. Percentages of DNA shared are also helpful in determining how, or if, common relatives are likely related through the same ancestors. I wish all of the DNA testing companies had these outstanding features.
 
 
 
 

 
 


Friday, February 10, 2017

How To Get The Most Out Of DNA Segment Data

Part 1) Why Segment Are Important And Tools For Working With Them

There are a number of free tools on the Internet which will allow you to visualize, in a graph format, and compare segments you share with your atDNA matches. This is important because the location, length, and overlap of segments with matches determines how likely a match shares a common ancestor within the genealogical time frame. Collecting overlapping, and shared segments, can help identify common ancestors even if a match hasn't posted a tree.

Family Tree DNA and 23andMe provide the all important segment data. AncestryDNA does not. If AncestryDNA would provide this important information we could solve more brickwalls with DNA.

Family Tree DNA and 23andMe provide chromosome browsers (a graphic representation of segments), which allow us to compare segments with up to 5 matches at a time. 23andMe allows us to compare match cousins not just with ourselves, but with our other matches. Family Tree DNA only allows comparison with matches who match ourselves, we can't compare match cousins with each other to see if they match one another, and by how much.

The downside of 23andMe is we can only compare our DNA segments with those who have agreed to share with us (blue dots), or have agreed to share with everyone (purple dot matches). For me that is 500 out of 1200 matches. Family Tree DNA allows us to compare with all of our matches, no agreement needed.

The chromosome browsers are useful, but it's often necessary to compare with more than 5 matches at a time. In this case there are online and downloadable tools which sort matches by chromosomes in an unlimited way. Sorting by chromosomes allows you to see overlapping and shared segments, which can help you to find common ancestors. Seeing all matches sorted that way allows us to find places, on each chromosome, where several people match each other. If they are on the same side, maternal or paternal, and share in the same place this is called triangulation.
Here are some tools which allow you to compare an unlimited number of matches sorted by segment in graph format:
  1. Genome Mate Pro: This is the best tool for segment comparison. It does require you to download and install it. It's free, but a donation would be appreciated. You can upload all of your matches from the Family Tree DNA Chromosome browser page. You can upload the aggregate file from 23andMe's DNA Relative page. You can also upload matches and segment data from GEDmatch. This tool does require you to study the manual, or watch YouTube videos to use. It's more complicated than the other tools. This is the only tool that allows you to compare matches from all of the databases making the complicated time consuming setup worthwhile.
  2. DNAGedcom: Description from site; "The Autosomal DNA Segment Analyzer (ADSA) is a tool that takes your data from Family Tree DNA or GEDMATCH and constructs tables that include match and segment information as well as a visual graph of overlapping segments..."
  3. Double Match Triangulator: This tool is only for Family Tree DNA. It also requires you to download it. Not as complicated to use as Genome Mate Pro, but the spreadsheet layout requires you to have a spreadsheet program installed, such as Excel. It's also not as easy to navigate and compare as Genome Mate.

Part 2) Working With Genome Mate Pro


This past week I've started from scratch at Genome Mate Pro. I lost some data when the program suddenly crashed. It's important to use the backup feature provided, in case it crashes. I decided to mark all chromosome segments according to which side the match is on,  i.e. maternal or paternal grandparent they are associated with.. To do this I first had to filter the matches by my mother's side or father's side.  Family Tree DNA and 23andMe provide that information if you have tested at least one parent (my mother tested).


To the right, in the snip above, is a filter allowing you to display matches on you maternal or paternal side, if you've tested at least one parent. Since I tested my mother the options are display matches on mother's side or not mother's side. What I did in this case was select matches on my mother's side then display Genome Mate Pro and the 23andMe windows side by side. I did the same for FTDNA.


I would look up each ancestor listed at 23andMe, or FTDNA, then mark them as M for maternal. (so far I've only finished my maternal line). I would then select the most recent common ancestor shared with this match. If I didn't know I would select the most likely grandparent the match was associated with. I was able to do this because my maternal grandmother was Nicaraguan, my maternal grandfather was Scots-Irish and German. 

 


Initially I assumed all Anglo surnames belonged to my Grandfather, and Hispanic names belonged to my maternal grandmother. At 23andMe I had an advantage. I could check the ethnic makeup of a match to see if they had the typical Nicaraguan ethnicity percentages. I could also check at both 23andMe and FTDNA to see if they had Nicaragua listed as a place of origin. Another way to verify I was attributing the correct matches to my grandmother was the check to see if they had common matches who did have Nicaragua named as a place of origin.

After marking your matches M or P and ascribing them to the most recent common ancestor the chromosome graph at the top of each chromosome page will begin to be colored in showing exactly where on the chromosome each segment appears. When I started marking out my maternal line I selected my maternal grandfather for all unknown relationship matches with Anglo surnames, and looked up all Hispanic surname matches, then attributed the Nicaraguan matches to my maternal grandmother. This actually didn't provide the color coded separation I was looking for so I went back and instead selected my grandmother's several times great-grandfather so I would get a clearly different shade of blue, because I wanted to distinguish the segments I received from her from those I got from my grandfather. The navy blue segments are segments my grandmother passed down to me, and the light blues and other colors are those from my grandfather. I have not been able to identify the common ancestor for any of my grandmother's segments, I only know these matches are Nicaraguan, that's why they are all navy blue.


I've started with my Maternal line. Paternal not yet colored. Dark blue grandmother light blue grandfather.

 
The light blue and dark blue should fit together like puzzle pieces, or be separated by gaps due to the fact not all my distant cousins have tested. They shouldn't overlap. This is generally the case. However, I have found a problem on one chromosome where my maternal grandfather and grandmother's segments overlap by about 10cM's. It could be the segment lengths are actually different than those provided as different companies segment lengths vary, or they may be within a false positive area. Genome Mate Pro will tell you if have a segment is in a false positive area. It will give you the exact location of this area if you move your mouse point over the ?, P, or M.

In the chart below we see the dark blues segments of my Nicaraguan grandmother's matches, and the other colors associated with my maternal grandfather, which are generally, separate as they should be. You can also see I've started marking out my father's match, which are on the top row of every chromosome. It's interesting to see where each of my grandparents' segments are. I'm guessing chromosome 17 is probably all from my grandmother, because out of hundreds of matches no one is showing up on this chromosome on the maternal side, and not as many Nicaraguans have tested. Chromosome 9 appears to be all from my maternal grandfather. So if I have a match on Chromosome 9 I can assume it's from my maternal grandfather.


I've overlaid Genome Mate's chromosome chart with 23andMe's ethnicity chromosome chart. The ethnicity chromosome chart shows segments color coded by ethnicity. When I overlaid both charts the ethnicities associated with Nicaragua, i.e. Southern European, Native American, and African do line up with my grandmother's navy blue segments. This suggests that the ethnicity results were accurate for her and I. You can see the X is mostly Native American aligning with my grandmother's ancestry, and our Nicaraguan matches on the X. Native American is the yellowish color.

Genome Mate segment map overlaid on 23andMe Ethnicity map

After marking all of my maternal grandparents matches I went through them chromosome by chromosome. I looked for overlapping segments. I then attempted to figure out which grandparent each match was associated with. I found some Anglo surnames segments overlapping with Nicaraguan matches. I then checked FTDNA and 23andMe for places of origin. I found one of these matches did have Nicaragua has a place of origin. I was able to place another Anglo surname match on my grandmother's Nicaraguan side based on the ethnicity results provided for this 23andMe match.


Going through matches chromosome by chromosome allowed me to identify more matches with common ancestors on my grandfather's side. This has been a very rewarding exercise as far as sorting matches by grandparent, and seeing exactly where on each chromosome each grandparent passed down segments to me. If you have ancestors from very different ethnic backgrounds an exercise like this can be very helpful. Even if your ancestors are all from the same ethnic background uploading to Genome Mate Pro and looking for shared and overlapping segments can help to identify how matches are related to you.