Triangulation and how to use it to further your research
Saturday 9 November 2019 10:00
am - 4:30 pm
WWW version:
YouTube version:
TBA
***** NB: FamilyTreeDNA kits will be
available after this talk for anyone interested via the DNA
Outreach IRL project *****
Outline
Introduction
There are many types of triangulation in genealogy and in the
wider
world.
- In the wider world:
- In genealogy:
- Sometimes the three angles in the triangle are two living
DNA matches
(cousins) and their most recent common ancestor (or, more often,
ancestral couple), sometimes referred to as tree triangulation.
- Sometimes the three angles in the triangle are three
living DNA matches
(cousins), sometimes referred to as segment
triangulation.
- In both cases, the objective is to identify the hitherto
unknown most
recent common ancestor (or, more often, ancestral couple) of the DNA
matches, and thus to prove that they are cousins, and to identify the
precise relationship(s) between them.
See the ISOGG Wiki for more on the origins
and usage of the word triangulation.
Shared,
or In Common With (ICW), matches
A group of three or more individuals who all meet the relevant
matching
criteria with each other are likely
to share a recent common ancestor (or,
more often, ancestral couple).
When I find a new match, I am usually anxious to identify
the most distant known ancestor through whom I am related to
the
new match.
The
matches that I share with the new match are usually the first clue to
solving this puzzle.
All
of the DNA
comparison websites allow one to identify the shared matches of
two individuals in
some form
or another.
The matching criteria vary from one DNA comparison website to
another.
The stricter the matching criteria, the more significant the
shared matches.
FTDNA
Family Finder
To find the shared matches of two individuals who match each
other:
- go to the match list of one of the
individuals
- tick the box opposite the other individual
- click the In Common With button on the 5th line from the
top of the
window
- matches are sorted by closeness of relationship to the
logged-in individual
- A can see that B matches C but cannot see the cM shared by
B and C.
- A's shared matches with B will be the same as B's shared
matches with A, but in a different order.
You will eventually identify a group of individuals, all of
whom you
suspect descend from a single common ancestor (or ancestral couple).
To see whether up to 10 individuals who match you also match
each other:
- Add them to the Selected Matches box on the Family Finder - Matrix page
- To find the desired surname in the Matches box, click on
any match and start typing the surname
- After finding the surname, ctrl-click on the desired
individuals
- Click the Add>> button to move all the
selected individuals to the Selected Matches box
To find the shared matches and shared cM of two or more individuals who
belong to the same project, whether or not they match each other:
AncestryDNA
On each match page, there is a Shared Matches link. (Screenshot.)
The Shared Matches are those with Shared DNA of 20 cM or more
with both individuals.
So C can appear in the shared matches of A and B even if B
does not appear in the shared matches of A and C (if C shares more than
20cM with A but B shares lesss than 20cM with A).
Matches are sorted by closeness of relationship to the
logged-in individual.
A can see that B matches C but cannot see the cM shared by B
and C.
MyHeritage
When the Review DNA Match page eventually
completes loading, the Shared DNA Matches section:
- reveals that "you share the following 1,532 DNA Matches"
- lists the top 10 shared matches
- allows further shared matches to be slowly loaded, 10 at a
time
- may demand money, depending on when you uploaded or what
subscriptions you have purchased
Matches are sorted by the sum of the centiMorgans shared
with the two individuals.
A can see not only that B matches C but also the cM shared by B and C.
A's shared matches with B will be exactly the same as B's shared
matches with A, in the same order.
GEDmatch
To find the shared matches of two individuals, whether or not
they match each
other, use the "People who match both, or 1 of 2 kits" tool on the many
menu.
This lists shared
matches of Kit 1 and Kit 2, no matter how much DNA or how little DNA
Kit 1 shares with
Kit 2.
Matches are sorted by closeness of relationship to Kit 1.
To sort by closeness of relationship to Kit 2, re-use the tool with the
kits in reverse order.
If you login in one browser tab, and open
the Multi Kit Analysis menu in another
tab (via a hyperlink or bookmark), then you can run an Autosomal Matrix
Comparison on up to 100 kits.
While the FTDNA user matrix shows only whether or not kits match, the
FTDNA administrator matrix and the GEDmatch matrix shows the shared
centiMorgans.
Triangulated
matches
Triangulation
and phasing
are
really opposite sides of the same coin. If V is
half-identical on the same region with W and Z, then there are two
possibilities:
- W and Z are half-identical to each other on this region, in
which case V, W and Z probably inherited an identical segment in this
region from a single common ancestor and the relationship can be
described as triangulated;
or
- W and Z are not
half-identical to each other on this region, in which case V is
probably
related to W on V's paternal side and V is probably related to Z on V's
maternal side, or vice
versa, and V's autosomal DNA in this region can be phased.
The ADSA tool by Don Worth at DNAGedcom
provides a graphical
representation of triangulation and phasing.
Triangulation
groups
The ultimate objective is to collect DNA matches into triangulation groups.
A triangulation group is a set of three or more people who are all
half-identical to each of the
other group members on overlapping regions.
The more individuals who are added to the triangulation group,
the smaller the overlap may become.
A triangulation group of three or more individuals are very likely to share
a recent common ancestor (or,
more often, ancestral couple).
Examples
and counter-examples
Example 1
9 106,881,785 124,347,952 24.0 4,938 Hincken/Morrissey
9 114,714,102 132,034,981 24.4 4,397 Quinn/Hincken
9 114,714,102 125,890,734 15.7 2,999 Morrissey/Quinn
All three members of the triangulation group are
half-identical to each other from 114,714,102 to 124,347,952.
The interpolator can be used to
confirm that the (sex-averaged) length of this overlap region is
approximately 13.7cM:
| Chr |
Query (bp) |
Sex-averaged (Kos cM) |
Female (Kos cM) |
Male (Kos cM) |
| 9 |
114714102 |
118.8510593879110 |
147.9192814455530 |
91.1703069169749 |
| 9 |
124347952 |
132.5099030861300 |
164.5840116062510 |
101.8401906721300 |
This region is long enough that it almost certainly contains a
DNA
segment inherited by all the members of the triangulation group from a
relatively recent single common ancestor.
A triangulation group including at least one
subgroup who are known relatives of each other is of particular value.
In this
example, Hincken and Quinn are known third cousins, with most recent
common ancestral couple Denis O'Connell (died 1887 aged 90) and Kate
O'Dea
(died 1889 aged 78).
It can then be concluded
that the
other members of the triangulation group are related to one of the most
recent common ancestral couple of the known relatives, or even
descended from both.
In this case, we can infer that Morrissey is
related to either Denis O'Connell or Kate O'Dea. The total amount
of shared DNA will indicate whether the most recent common ancestor of
the entire triangulation group is upstream or downstream from the known
most common ancestral couple of the known relatives. In this case,
Morrissey has known ancestry from the same townland as the others, so
geographic evidence added to genetic evidence points in a very specific
direction.
Example 2
8 8,292,285 54,555,905 50.2 10692 Claire/Ed
8 17,609,398 38,378,638 27.4 5563 Ed/Michael
8 17,648,866 37,812,773 26.9 5499 Michael/Claire
Claire and Ed are fourth cousins, descended from two O'Dea sisters, Mrs
Lynch and Mrs
O'Connell respectively, who lived in the townland of Moveen West,
County Clare..
So their 50.2 centiMorgan half-identical region must contain an O'Dea
segment.
So Michael's grandfather Eugene Lynch, born in "Ireland" must
be related to Mr O'Dea
(first name still unknown) or Mrs O'Dea (Neans) or descend from both.
Eugene and Michael's surname is Lynch, so they must surely descend from
Michael
Lynch and Mary O'Dea.
Eugene (41 in 1910, migrated 1887) is too young to be Michael and
Mary's son (their known children have implied birth years 1824-1841).
Eugene's descendants share too much DNA with other descendants of
Michael Lynch and Mary O'Dea for him to be their greatgrandson.
So he must be their grandson.
Michael has now gone from knowing that his ancestors came from
"Ireland" to knowing the precise townland where his ancestors lived.
As half-identical regions have fuzzy
boundaries, it is rare to find
that even two of the half-identical regions in a triangulation group
share either the
same beginning or ending location. Between the two examples above,
there is just
one boundary location (114,714,102)
common to two of the three half-identical regions. One should
concentrate on looking for
overlaps rather than exact shared boundaries, which have no special
significance.
If a triangulation group contains two subgroups of known
relatives, then it can be concluded that one of the most recent common
ancestral couple of the first subgroup is related to one of the most
recent common ancestral couple of the second subgroup.
Counter-example: the Smith, Parker and Keas siblings
There is an exception to every rule.
It is possible to find triangulated matches without a common
ancestor.
- 14 September 1831
- Thomas Parker married Mary Keas (my GGGgrandparents);
- 3 Mar
1840
- Francis Parker (Thomas's brother) married Margaret Smith;
- 26 May 1841
- Joseph Smith (Margaret's brother) married Ellen Keas
(Mary's sister).
They produced three families, each of which were
first cousins to the other two, but who didn't have a single common
ancestral couple!
Consider:
- Catherine Parker, daughter of the first couple;
- Anne Parker, daughter of the second couple;
- George Smith, son of the third couple.
Each of these, as first
cousins, shared many half-identical regions with
both of the others, probably including some regions in which:
- the two
Parkers
had an identical segment from a Parker grandparent,
- one of the Parkers
and the
Smith had an identical segment from a Smith grandparent, and
- the
other Parker and the Smith had an identical segment from a Keas
grandparent.
In this example, the three cousins
will all "match" each other genetically.
They are expected to triangulate on 1/64th of their autosomal DNA, so
(like a comparison between two third cousins) there is about a 90%
chance of finding a triangulated match.
On closer examination it
will be found that there is no common ancestral couple of all three.
[To complicate matters further, Anne Parker went on to marry
her first cousin George Smith!]
With this caveat, the triangulated matches that I share with a
new match are
usually the second
clue to identifying through which of my most distant
known ancestors I am related to the new match.
Some of the
DNA
comparison websites allow one to identify the triangulated matches
of two individuals
in some form
or another.
FTDNA
Family Finder
One has to be a little devious to find triangulated matches
directly at
FTDNA.
The Linked Relationship feature is
designed to identify matches who
triangulate with known third cousins or closer, and then dump them all
together again in paternal and maternal buckets.
It can
alternatively be used to identify all the other matches who triangulate
with a particular match of interest.
To do this, you will want to have
just one linked relationship at a time on one side, say Paternal. You
can just link the match of
interest temporarily as your father, whom you should mark as
living,
to avoid any chances that your matches or
project administrators interpret this device as proof of a relationship.
Family matching will then put all the
matches who triangulate with you and the match of interest in the
Paternal bucket. You can still use the Maternal bucket in the normal
way to collect your triangulated maternal matches (or you can use a
maternal ancestor in this procedure and use the Paternal bucket in the
normal way).
When you have recorded the list of triangulated matches, you should
probably reverse the process by returning to the myFTDNA - myFamilyTree
tab and deleting the individual from the tree.
Then you can repeat the process for any other match of interest.
If you want to use the Linked Relationships feature both for
bucketing
both Paternal and Maternal matches and for identifying triangulated
matches
with individual matches, and if your matches include known third
cousins or closer on both paternal and maternal sides, then you may
want to have two Family Finder
kits, e.g. a kit based on swabs sent to FTDNA with a full pedigree
chart for paternal/maternal phasing; and a kit based on an autosomal
transfer from another laboratory
with a minimal pedigree chart for
identifying triangulated
matches.
AncestryDNA
AncestryDNA refuses to provide any way of identifying triangulated
matches.
MyHeritage
MyHeritage shared match lists include a
symbol identifying
which of these matches are triangulated.
There appears to be no way to filter the list of shared
matches to show only the triangulated matches.
GEDmatch
To use the triangulation tools, you are expected to subscribe
to Tier 1 (USD10 for one month).
If you login in one browser tab, then you
can open
these Tier 1 tools in another tab:
- the Multi Kit Analysis menu allows you
to search for triangulations involving Kit 1 and any two
or more of Kits 2,3,4,...
- the Triangulation
tool allows you to find all triangulation groups for the selected kit
at the selected thresholds
- the Segment Search tool allows more
flexible manual investigation of phasing and triangulation
I use a Microsoft Excel macro to tidy and save the Segment Search
output.