DMT's Method of Chromosome Mapping

 
Double Match Triangulator tries to get the most information possible out of your DNA segment matches. It uses techniques that include DNA painting as done at Jonny Perl's website dnapainter.com and clustering as started by Dana Leeds and enhanced by others. DMT includes the ideas and concepts of Tim Janzen and Jim Bartlett who use triangulation groups to map chromosomes to ancestors. Tim and Jim did this manually for years. Tim often presents his techniques at genealogy conferences and online. Jim created a website www.segmentology.org  that described his methods.
 
Double Match Triangulator does the work to automate these methods the way a person would manually do it if they had all the data and enough time and patience.
 
DMT uses a primary person whose DNA is of most interest, a second person whose matches are compared to the primary person, and the results will include the matches ofall the people who the primary person matches to.
 
  • Person A is the person of interest whose chromosomes you are mapping. 
  • Person B is the person who you have a match file for who is really only useful if they match Person A somewhere. It is helpful to know the common ancestor between Person A and Person B, but that is not necessary.
  • Person C refers to any of the people who Person A matches to. Knowing the common ancestor of some of the C people is what will allow you to map your chromosomes to your ancestors.
     
    Double Match Triangulator can be run without knowing any common ancestors. In that case, all mappings will be to "U" (unknown), but you will still be shown all the triangulations that Person A and B share, plus other double matches and single matches of Person A.
     
These are the elements of chromosome mapping as done by Double Match Triangulator:
 
  • Triangulations:  provide additional information to map additional segments using the fact that there are only two triangulation groups (one paternal, one maternal) at any one position on a chromosome, and that one triangulation group must comes from a single ancestral path. Triangulations also allow the use of smaller segments than single matches, which according to Jim Bartlett allow you to have confidence down to at least 7 cM and maybe even 5 cM, compared to the 15 cM reliability level for single matches.
     
  • Missing A-B matches:  This is where Person A matches Person C, and Person B matches Person C on the same segment, but Person A does not match Person B on that segment. This is not a triangulation and indicates that Person B is most likely matching Person C through a different relationship than Person A has with Person C.
     
  • Missing B-C matches:  This is where Person A matches Person C, and Person A matches Person B on the same segment, but Person B does not match Person C on that segment. This is not much better than a single A-C match, except that it does indicate that Person A matches Person B, so Person B is most likely matching A on the opposite parent that Person C is matching.
     
  • Single A-C matches: If neither B matches C, nor A matches B on this segment, then no additional information is available from this match. If the match is above a user threshold (the default being 15 cM) then the match will still be included in the analysis because it is most likely a real match. It will be assumed to not be matching the chromosome of the parent that has triangulations on that segment.
     
  • Inferred matches:  Information may be available when Person B matches Person C on a segment that Person A does not match Person C. In some cases, a technique named "Inferred Matching" by Jonny Perl can be used. Inferred matching requires that both the common ancestor between Person A and Person B and the common ancestor between Person A and Person C are known. It tells you that the segment where Person B matches Person C cannot be from a specific common ancestor. Because if it was, then Person A and Person C would match as well.
     
  • MRCAs and Ancestral Paths: DMT uses a notation to represent the Most Recent Common Ancestor (MRCA) of Person C with Person A and almost the same notation for ancestral paths. The notation is simply a string of "F"s and "M"s. So an ancestral path through Person A's father's mother's mother would be FMM. The only difference in notation is that an MRCA can end with an "R" which represents a paiR of paRents who both Person A and Person C share. The common ancestor is always from Person A's point of view. If Person C is a second cousin through Person A's father's mother's parents, then the MRCA would be FMR. There would be two possible ancestral paths shared with the cousin:  FMM and FMF. Siblings, nephews, nieces and children of Person A will all have an MRCA of just R. Half-relatives would need to go up one level to get to the R, for example, a half-sibling would be either FR or MR depending on whether the common parent is father or mother.
     
 

How DMT Does It

 
Double Match Triangulator automates the steps a person would take using double matching to map their chromosomes. These are the steps DMT follows.
 
  • Eliminate Small Segments - Double Match Triangulator uses the "Min Triang" value to exclude small segments from analysis. The DMT default is 7 cM. Triangulations less than 7 cM might be matches by chance (matching on more than one chromosome) and thus be false triangulations. Jim Bartlett believes this number might be as low as 5 cM.
     
  • Double Matches - All the matches in Person A's file and Person B's file are compared. Every Person C who matches both Person A and Person B on an overlapping segment on the same chromosome is a double match. It's called a double match because there is both an A-C match and a B-C match.
     
  • Triangulating Segments - Every double match is checked to see if there is also an A-B match on the same segment. This will find all triangulating segments that Person A and Person B have with any Person C. Double Match Triangulator finds and displays every triangulating segment that is contained in the segment data after the small segments are removed. For each triangulating segment where the MRCA of Person C is supplied, that MRCA is "painted" onto the chromosome for that parent. It is painted for the entire length of the A-C match while marking which sections have triangulations.
     
  • Missing AB Matches - If there is no A-B match on the double match, then it is called a Missing AB match. This is not a triangulation which usually indicates that this was not a segment passed down from a common ancestor to Persons A, B and C.  We are saying specifically here, that A does not match B on the segment. We want to be sure, so the A-C match must be at least the Min Single value (default 15 cM) to ensure that we are not dealing with a possible false match. For each Missing AB match where the MRCA of Person C is supplied, that MRCA is "painted" onto the chromosome for that parent for the length of the A-C match less 1.5 Mbp on either end to allow for a bit of random matching.
     
  • Missing BC Matches - If there is no double match, but A matches B on the segment, then it is called a Missing BC match. This is also not a triangulation. However, if B is in common with C on other segments, then B and C are confirmed not to match on this segment. Like Missing AB matches, the A-C match must be at least the Min Single value. For each Missing BC match where the MRCA of Person C is supplied, that MRCA is "painted" onto the chromosome for the parent for the length of the A-C match less 1.5 Mbp on either end to allow for a bit of random matching.
     
  • Inferred Matches - When Person B matches Person C on a segment, but Person A does not match Person C on that segment, then it may be possible to infer something about Person A's ancestral line along the same segment. Depending on how Person A is related to Person B, the fact that B is matching may indicate that A cannot match along some related ancestral line. To infer anything, you must know both Person A's MRCA with Person B and MRCA with Person C. This technique works very well when Person B or C is a sibling or niece/nephew, because you can often eliminate a whole grandparent from Person A's possible ancestral lines on a segment, meaning Person A's ancestral line must go through the other grandparent on that segment. I wrote a blog post about Inferred Segment Matches in March 2019. Since then, I determined all the ways that inferred matches can give results. Anyone descended from both of Person A's parents are able to refute a grandparent. For other people, the refutation is 2 generations beyond the closer MRCA up the line of the further MRCA, meaning one MRCA must be at least 2 generations past the other. Double Match Triangulator finds, displays and makes use of every inferred match that is at least of Min Single size that is contained in the segment data.
     
  • Single Matches - Any A-C matches that did not triangulate and were not Missing A-B or Missing B-C matches are now considered. DMT will "paint" the MRCA of any match with centimorgans of at least "Min Single" (default 15 cM) onto the parental chromosome and will be designated as "Single AC". If the single match has less cM than "Min Single" but more than "Min Triang", DMT will call it "Small AC" and will display it in an "Unassigned" section to emphasize that this segment doesn't triangulate and may be by chance.
     
  • Ancestral Paths for Every Mbp - DMT next takes the "paintings" done above, and comes up with a consensus for the ancestral path at every Mega Base Pair (Mbp) for each parent along each chromosome. Inferred matches are taken into account and sometimes will determine the ancestral path when the MRCAs themselves do not. A "majority rules" type of consensus is used to first determine the grandparent, then the great-grandparent, etc., until we've reached the limit of the MRCAs that have been assigned.
     
  • Triangulation Groups - Triangulation group boundaries are determined for the father and mother chromosomes based on the number of matches that start and stop at each Mbp. All segments assigned to a parent, whether triangulating or not, are then assigned a triangulation group based on the boundaries that will contain them.
     
  • Grandparent Extensions - All Mbp within each triangulation groups should come from the same grandparent. If a grandparent is assigned to some but not all Mbp in a triangulation group, then the same grandparent can be assigned to the other Mbp in the group. If both grandparents are each assigned to some Mbp in a triangulation group, then the grandparent with the most assignments is used and the others are overwritten.
     
  • Ancestral Paths for Every Segment Match - DMT next assigns an ancestral path to every segment. It first sees if it can determine the parental chromosome of the segment. If there were triangulations painted on the Mbp making up the segment, then DMT sees which parent the triangulations were painted on. If they were painted on just one parent, then triangulations must be on that parent and other matches (Missing AB, Missing BC and Single Matches) must be on the other parent. If triangulations were painted on both parents (which can validly occur if Person B matches Person A on both parents as often happens with siblings), then the parental side is not known. If the parental side of the segment is not known but the person has an MRCA assigned, then then the parental side of the MRCA is used. If a parental side is determined, then the ancestral path at each Mbp for that parent is found, and a consensus is determined and assigned to the segment match. When no parental side is determined, then the segment is put into the "U" (unassigned) group..
     
  • Clustering People - Each person now has all their segments assigned. Most of one person's segments are often on one ancestral path, but they still might have some segments on different paths or on unassigned paths. They may be related through both of Person A's parents and match on both father's side and mother's side. DMT now goes through each person and determines what the majority of the ancestral paths agree on. The resulting path will be assigned as that person's Cluster. This cluster is likely the ancestral path through which this person is most closely related to Person A.
     
  • This completes the chromosome mapping process for two segment match files between Person A and one Person B. DMT produces an Excel file with two pages containing the results. The Chromosome Map page lists all the A-C matches along with information about each match ordered by triangulation group. The People page lists all the people who match Person A ordered by cluster along with information about that person's segment matches. A log file summarizes the run, and a comma delimited csv file is optionally created with triangulation group information in a format suitable for upload to DNA Painter.
     
  • If the "Folder B" option was selected, then steps 1 through 13 will be done for File A against each match file in Folder B.
     
  • If the "Combine all results" option was selected, then steps 1 through 13 will be done for File A against each match file in Folder B, and then the results will be combined. DMT produces 23 Chromosome files and creates or updates Person A's People file. A log file summarizes the run, and a comma delimited csv file is optionally created with triangulation group information in a format suitable for upload to DNA Painter.