Ideas, Tips and Tricks

 

 

Double Match Triangulator will provide you with ALL triangulations that Person A has with Person B down to the "Min Triang" limit that you specify.

 

Double Match Triangulator will also provide you with ALL inferred matches that Person A has with Person B down to the "Min Single" limit. Inferred matches can only be determined from a Person B and Person C who both have their MRCAs assigned.

 

Double Match Triangulator will also take your known MRCAs and automate the DNA Painting procedure that you would do it you had the time and patience to paint thousands of segment. It will include the use of triangulation information which most DNA Painting methods do not consider. And it will optionally produce DNA Painter output file that can be uploaded to Jonny Perl's Dna Painter tool.

 

Double Match Triangulator will organize the people into Clusters that correspond to your MRCAs, providing different and additional information from other clustering tools.
 

 

 

Double Match Triangulator is designed to help you extract information from your DNA matches. It automates a lot of the work of comparing segments and assigning ancestral paths that you would want to do.

 

The results are only as good as the segment data and your MRCA assignments. Generally segment data is more "correct" than "incorrect" and the same hopefully can be said for your MRCA assignments. What that means is that DMT's results should generally be good, but are definitely subject to error. You should check any results you use to see that they seem logical and make sense to you.

 

Think of DMT like a clustering algorithm that does its best job to put people together into clusters that make them most similar. That in no means gives you perfect results. It just gives a best-possible algorithmic estimate of what the most likely results are. That is good information that can help you to place your DNA relatives into your family tree.

 

 

 

If you assign an MRCA to a person, but the person is also related another way, then the segments through the other relationship will be assigned an incorrect ancestral path. Since DMT uses a consensus-based decision as to what the ancestral path of a segment is, hopefully is most cases, correct assignments will outnumber incorrect assignments and the final mapping will be correct. But not always.

 

If a segment match is a match by chance or false match, the MRCA will be assigned if the match is included. You can manipulate the segments that will or will not be triangulated and included with the Min Triang and Min Segment settings.

 

 

 

Any segment match file can only include people who tested before you downloaded the file. A newer file will have people who tested that the older file did not. Obviously, you would like the newest file possible, but maybe you got the segment match file a while ago from someone else and have no way of updating it. Can you still use that file?

 

If you compare a newer file as File A with an older file as File B, the newer File A may match Person C, but the older File B may not because Person C did not test yet. That match could actually be a triangulation, but would then only show up as a single A-C match. If Person A has any triangulations with other people on the segment, then the A-C match will be incorrectly assigned to the opposite parent.

 

If you compare an older file as File A with a newer file as File B, then the Person A file will not contain any newer matches and there should be no problems with triangulations.

 

So the general rule if you have a choice, is use a File A that is older than any of your File B's.  If that's not possible, then just be aware that for C people who tested after B tested may have some segments assigned by DMT to the opposite parent when they should be triangulations.

 

 

 

I create a DNA folder for all my DNA analysis.

 

Under that, I have a folder for FTDNA, one for 23andMe, one for MyHeritage and one for GEDmatch.

 

Under those, I have a folder for each person who I got the match files from.  My own files and the files of people I administer are in a folder called "Louis Kessler".  For GEDmatch, you can download files yourself, so you could put them all in the GEDmatch folder if you want.

 

Then, I can run myself (or anyone else) as File A against all the files I got from a specific person, or all my matches by specifying the DNA company folder and checking

 

 

 

DMT only can compare same to same. i.e. FTDNA to FTDNA, 23andMe to 23andMe, MyHeritage to MyHeritage, and GEDmatch to GEDmatch. That is mainly because the same people at different companies very often have different names.

 

If you have data for a Person A from multiple companies, then what you can do is use is set up DMT to do a Combine all Results run and do that once for each company. Each company will have a different People file for Person A, and you'll have to enter Person A's MRCAs into each of them. But then you'll be able to compare the result files between the companies yourself, or create DNA Painter files and load them into DNA Painter and use the DNA Painter tools to compare them.

 

Which company? All of them that you have data for, of course. Every one will give different information because you match to different people at the different companies. Ancestry DNA unfortunately doesn't supply its users with their match data, so the only way to include an Ancestry DNA tester is if they've uploaded one of the other companies.

 

 

 

If you don't know any of your DNA match's MRCAs, then DMT won't be able to calculate any ancestral paths for you. The ancestral paths can only be calculated as deep as your most distant MRCA is. So if you only know a second cousin, e.g. whose MRCA is MFR, then usually the best DMT can do for you is to separate out MF from MM. The more MRCAs you know exactly, the better.

 

However, do realize that an MRCA very deep, e.g. MMFFFMR, might be a common ancestor, but it might not be the only common ancestor and it might not be the Most Recent Common Ancestor.  Even so, it is definitely worth trying it. If that person is only related through that ancestral path, then you got it right.

 

To test if your deep path might is likely the best, try a run leaving its MRCA blank. See what ancestral paths DMT assigns its segments and what cluster the person is put into. If the cluster and most of the segments agree with the deep path, then you are likely safe to use that deep path.

 

 

 

If you don't know the full MRCA of a DNA match, you can leave it blank and DMT will do its best to assign ancestral paths to all the segment matches without that information.

 

But if you do know the match is along a specific line somewhere, feel free to enter the known part of the MRCA in the people file. Partial MRCAs are indicated by not having a trailing R. That means the true MRCA is at least as long and may be longer than what is given but the rest is unknown. e.g. you can use MM if you know the person is related somehow through your Mother's Mother's line. The Leeds Method and various clustering tools may help you assign partial MRCAs to some of your matches, especially at the grandparent level.

 

DMT will use full or partial MRCAs to be each segment's assumed starting ancestral path.

 

If you have a relative you know is related more than one way, then some segments will come from one MRCA and some will come from another. You can only pick one MRCA, so segments from the other will be assigned incorrectly. You can leave the MRCA blank, or you can use an partial MRCA (one without an "R" at the end) if the two MRCAs start along the same path.

 

 

 

A male only has one chromosome 23 (aka X) that is passed from his mother. DMT uses the sex you enter for Person A. If Person A is a male, DMT will calculate all ancestral paths on his X chromosome as coming only from his mother.

 

Any X matches with someone who has FF as part of their MRCA will be considered false. The X chromosome does not pass through two males.

 

 

 

Double Match Triangulator bases its triangulation groups boundaries on Mbp that have a local minimum number of triangulations running through that location. This means that there are some triangulations that end before this point and others that start after this point. This indicates a possible subdivision point that divides a triangulation group for a closer MRCA into two triangulation groups of more distant MRCAs.

 

The starting and ending locations of triangulation groups cannot be determined exactly because there can be by-chance matching at both the start and end of any real match. So the best that can be done are approximate boundaries between triangulation groups. DMT rounds triangulation group boundaries to the nearest Mbp, and even so, you'll see matches often extending a few Mbp past their boundary.

A key to interpreting matches and triangulation groups is that the start and end of any match is due to a recombination that occurred somewhere down from the Most Recent Common Ancestor, and that could have happened on the way down to Person A, down to Person B, or down to Person C. The Start-AC, End-AC, Start-BC and End-BC positions on the Chromosome Map page can help you identify whose line "owns" the endpoint. If that endpoint was owned by Person A, then the segment on the other side of the endpoint would have come from a different ancestor.
 

 

Be very careful on interpreting small segments. Single matches less than 15 cM might be matches by chance.  Triangulations under 7 cM might be matches by chance.

 

Generally, you should leave the DMT defaults at these 7 cM and 15 cM levels:

 

You're much better off trying to figure out new MRCAs to add, or getting other people's match files, than lowering these limits.

 

 

DMT has to use the data it is given. It uses all the matches from the segment match files filtered down to the Min Triang level and it uses the MRCAs that you provide. Often the data conflicts, e.g. a segment triangulates with someone on your father's father's side and someone else on your father's mother's side.

 

There are many possible reasons for these type of seemingly impossible results. Segment data isn't perfect. Company matching algorithms aren't perfect. A match may be by chance. A smaller segment may be filtered out. Your MRCA may not be correct. Your relative may be related more than one way. A triangulation may be through opposite parents. One person may triangulate with everyone else because their two chromosomes match with the one segment. Two different people might have the same name.

 

DMT attempts to take every assertion that your data makes and combine it into a consensus-based best estimate of what's going on. It uses human-like logic to "paint" the segments, assign ancestral paths, and cluster your matches. It does what you would want to do with the data, but does it in seconds rather than years.

 

One method you can use for analyzing segments matches and resolving conflicts is to produce both a "Combine all results" run and then uncheck it and produce all the individual one-to-one runs. First look through the combined results for anything that looks strange, Then check the relevant one-to-one runs to determine the details. You might want to look at the specific segment match files to see the raw data that was used. Almost always, you'll find that there is conflicting data, and the program is really doing the best it can with what it has.

 

If there is a specific person who is problematic, you can remove their MRCA, or use a partial MRCA, e.g. remove the "R" or reduce the MRCA to just "F" or "M" without an "R" a the the end and let DMT tell you what it suggests as the ancestral path of the person's segments and the cluster of the person.

 

You may also have some slightly-larger-than-small false segments. You may want to try increasing Min Triang and/or Min Single by a few cM.

 

Try the other tools, including DNA Painter, Leeds Method, Clustering, etc. and compare results.

 

The best ways to improve overall results is to find out how you're related to more people and add their MRCA. You can also get more segment match files for some of your matches, especially second, third and fourth cousins who will help identify the ancestral paths along their lines. At GEDmatch and 23andMe with DNAGedcom, you can obtain segment match files yourself for any of your matches. At Family Tree DNA, MyHeritage or 23andMe without DNAGedcom, you will have to ask your matches to send you their segment match files.

 

 

 

I will blog from time to time about DMT and how to make use of it. A list of my DMT-related blog posts that I try to keep up to date is at:

www.doublematchtriangulator.com/blog.php