Important!
You need to have Java Runtime Environment 1.5.0 or better to view this applet.
If the Java installation on your computer is not up-to-date,
then use
this link
to get a free download of the latest version of the JRE, Standard Edition
from Sun Microsystems.
Select checkboxes to activate the tools.
Notes:
The applet display is driven from an underlying GEDCOM file,
which contains all the data. Right now I mostly have only
lineage information for Aaron Stark descendants, but I am hoping
to add data for Archibald Stark descendants and other Stark lines
as well.
When the applet starts it computes a sort ordering for the
participants that puts related participants nearer to each
other, making it easier to identify family relationships.
It also computes a coloring that tries to give different colors
to unrelated participants. The coloring is used in the various
display windows.
By selecting the checkboxes, you can pop up tools that display
the Y-chromosome DNA data for the
participants in various ways, and that permit you to browse the
family trees of the participants.
The windows are connected in the sense that if you click on
an individual in one of the displays, the others will follow along.
Currently, the supported tools are:
Participant table:
Displays a table showing the names of the project participants,
their most distant known ancestor, and contact information.
Lineage browser:
Allows you to browse the family tree of the various participants,
displaying genealogy data in textual or graphical form.
Genetic distance display:
Displays a table showing the genetic distances between the
project participants, with close relationships highlighted.
The values displayed in the table are obtained by dividing
the genetic distance between the haplotypes of each pair
of individuals by the number of markers compared.
Values of less than 0.08 are highlighted.
At the moment you can see five distinct family groups emerging.
One of these is formed by descendants of Aaron Stark of Connecticut.
Another is formed by descendants of Archibald Stark of New Hampshire,
James Stark of Virginia, and Richard Starke of Virginia.
The exact relationship between Archibald and the Virginia
Starks is apparently not a matter of agreement among Stark
family researchers, although one can readily find Web sites
that claim various relationships (without citing sources!).
The third group is formed by descendants of Thomas Starke
of London, England.
The fourth group consists of descendants of Zerubabel Starks
of Tennessee/Virginia.
Two other closely related individuals are descended from
Thomas Starke (b. 1724) of Virginia.
Allele data:
This is the raw genetic data. Each row displays a
Y-chromosomal haplotype of one of the participants.
The columns correspond to marker sites, and are labeled
with standard names as used by FamilyTreeDNA.
The entries in a column are the allele values exhibited for
that marker by each of the participants.
An entry is highlighted for an individual if there exists
some other individual who on the basis of genetic distance
is likely to be related to the given individual but who has
a different value for this particular marker.
Thus, the highlighted entries cover all the genetic variation
within groups of related individuals.
Mutational models:
This screen computes minimum mutational models for the set of
all descendants of a selected "root" individual. Briefly, a
"mutational model" for a set of individuals is an assignment of
haplotypes to each of the individuals in the set that is
consistent with any known haplotypes for those individuals.
If we assume a certain "family tree" for the individuals,
then we can count the total number of mutations required to
"explain" the model in the context of that family tree.
A "minimum mutational model" for a family tree is an assignment
of haplotypes that requires the fewest mutations to explain
in the context of that family tree. Minimum mutational models
always exist if we are willing to consider arbitrarily large
numbers of mutations, however there may be more than one
minimum mutational model in a given situation.
The algorithm I implemented to find minimum mutational models
is similar to the algorithm originally due to Fitch and analyzed
by J. A. Hartigan
(Minimum Mutation Fits to a Given Tree,
Biometrics, Vol. 29,
No. 1. [Mar., 1973], pp. 53-65).
This algorithm is not efficient in the theoretical sense.
For efficient algorithms and a more modern discussion of
minimum mutational models and other similar ideas,
see
M. Vingron, J. Stoye, and Hannes Luz,
Algorithms for Phylogenetic Reconstructions
(Lecture Notes and Exercises, Winter 2002/2003).
I have not yet gotten around to incorporating into this tool
the efficient algorithms they discuss.
To use the tool on this screen, first use the lineage browser
to select the desired root individual. The descendants of this
individual will be shown on the Mutational Models screen,
together with known haplotype data for any markers on which
there is any disagreement between the selected descendants.
(If no markers are shown, there is no disagreement, and nothing
interesting to compute.) Click on "Run" to begin the calculation
of minimal mutational models. The search proceeds marker by
marker, with the marker currently under consideration being
indicated by a pink bar. When the calculation for a particular
marker has completed, allele values are placed in the column
for that marker. Yellow cells indicate allele values that
represent mutations between an individual and his father.
The calculation can be stopped and resumed using the "Stop"
and "Run" buttons. If there is more than one minimum mutational
model found, the "Select" menu can be used to choose the one
to be displayed.
Once calculations have been initiated by pressing "Run", the
root individual becomes locked so that selections on other
screens do not affect the contents of this screen.
In this case, to calculate models for another individual, it is
first necessary to click on "Unlock" before selecting the
new individual using the lineage browser.