Genetic Networks, Part 5: Use Ancestry Pro Tools’ Enhanced Shared Matches

In the previous blog posts, I defined genetic networks (Part 1), discussed the inclusion and exclusion criteria for matches within the network (Part 2), described how to find them (Part 3), and demonstrated how to efficiently review your matches’ family trees (Part 4). Now, I describe how to use Ancestry Pro Tools’ Enhanced Shared Matches feature to allocate member matches into sub-clusters of unlinked family groups. This can be accomplished even when these matches do not have family trees.

I’ve previously written about Ancestry’s Enhanced Shared Matches feature, but to quickly recap, the tool provides additional information about your matches’ matches. So, if you’re reviewing a DNA match and then click on the “Shared Matches” filter button, the Enhanced Shared Matches tool now shows how much DNA the match shares with your other matches.

As in prior posts, I labeled genetic network member matches as “A” (test taker), “B” (match being viewed), and “C” (matches both A and B, i.e., shared matches). Prior to the introduction of Enhanced Shared Matches, we could only see the amount of shared DNA (in cM) between “A” and “B” and “A” and “C”, but now we can also see how much DNA is shared between “B” and “C”, which can help us determine how the “C” match is potentially related to other matches. In the image below, we can see that because “C” shares 229 cM with “B”, “C” is more closely related to “B”, which the Shared cM Project suggests could be B’s second cousin.

Research Objective

To contextualize my use of Ancestry’s Enhanced Shared Matches feature with genetic networks, I start with a research objective. While my ultimate goal is to find one or both parents of my 4x great grandfather, William Hill (1775-1836), my intermediate objective with respect to genetic networks is to:

Identify relevant genetic networks for William Hill’s (1775-1836) ancestors to guide my documentary research efforts.

Genetic Network Strategy

Because I want to demonstrate the Enhanced Shared Matches feature, I use autosomal DNA matches within Ancestry. I’m using a cousin’s DNA matches because he is one generation closer to William Hill potentially having more original DNA of our shared ancestor than I do. Also, based on my assessment of his matches, he in fact does have more “Hill” matches than I do.

As discussed in Part 3 of the Genetic Network series, I used the EGGOS Search Strategy, which is a strategic tree triangulation method to isolate the shared matches to those only representing the ancestor of interest. I am fortunate in that my 4x great grandfather William Hill had children with multiple wives. My cousin, whose DNA matches I am using, is the “A” match (purple line in th image below) and descends through William Hill’s daughter Susan with his first wife Elizabeth Winland. The “B” match (pink) descends through William Hill’s daughter Matilda with his third wife Susan Whitman. Among all matches descending from William Hill and Susan Whitman, I selected a “B” match that had the highest amount of shared DNA in cM.

Because the only common ancestor between “A” and “B” is William Hill, I am able to isolate the shared “C” matches to William’s ancestry. If I would have used a “B” match descending from another child of William Hill and his first wife, then the resulting shared “C” matches would represent both Hill ancestry and Winland ancestry.

Selecting the Shared Matches filter when reviewing the “B” match identified 105 shared matches (C) between my cousin (A) and the descendant of Matilda Hill (B). After considerable effort in evaluating the matches’ family trees and building out other matches’ family trees, I was able to construct a genetic network by triangulating their trees (see below).

As depicted above in blue, I identified 30 matches descending through four of William Hill’s children. The observation that identified matches descend through multiple children suggest I identified the relevant ancestor or ancestral couple for these matches.

I also identified four other sub-clusters within the genetic network. One sub-cluster of 18 matches descending through two different children of Joseph M. Keel (green), one sub-cluster of 11 matches descending through two different children of Asa Linn (orange), and one sub-cluster of 7 matches descending through three children of Clay Harris (brown). A final sub-cluster of 20 matches was found for a group of Clark descendants descending through John Clark and/or Hezekiah Clark, who are probable brothers (yellow). The Keel, Linn, Harris, and Clark sub-clusters represent unlinked family clusters that I cannot yet connect with my Hill ancestral line but will be explored in Part 7 of this blog series (scheduled for release on October 21).

Of the remaining 19 of the 105 originally identified shared matches, four were misclassified genetic network matches (see Part 2 of this series), and 15 matches represent a group where I am unable to determine a connection to these other sub-clusters or among themselves (pink).

While I can’t discount the traditional genealogy I undertook to create the sub-clusters, the Ancestry Pro Tools’ Enhanced Shared Matches feature played an important role in reducing the amount of effort to create the sub-clusters and in establishing confidence for the observed sub-clusters.

Enhanced Shared Matches Tool

Expanded Pool of Matches
Of the 105 matches, 75 matches shared 20 cM or more with my cousin. Prior to subscribing to Enhanced Shared Matches, Ancestry would only present these higher confidence matches. Now, with Enhanced Shared Matches, I am able to see all shared matches down to 8 cM, which in this Hill genetic network produced an additional 30 matches.

Enhanced Shared Matches Benefit. The more matches we have, the greater the chance in finding the one critical match helping us identify the next generation. More matches also builds critical mass around unlinked family clusters within the genetic network assisting with prioritization of documentary research.

Despite having access to matches down to 8 cM, I place greater weight on the matches at or above 20 cM but also evaluate those below 20 cM. As I’ve written previously, small DNA matches can act like a compass and point our research in the direction toward sub-clusters within the genetic network that may be the most promising in producing results. That said, there is also a greater incidence of misclassified genetic network matches (see Part 2 of this series) and matches that are difficult to group into an unlinked family sub-cluster because they are perhaps too distantly related. The table below summarizes the match counts by sub-cluster and by those obtained without Enhanced Shared Matches and the additional ones gained by it.

Tree Building by Association
Using Forest Management principles discussed in Part 4 of this series, I evaluated the matches with family trees and constructed family trees for those with incomplete trees. This was a total of 64 matches. However, nearly a third of the matches (31 in total) had no trees, but this is where the Enhanced Shared Matches tool can help.

For any of the shared “C” matches without a tree, select that match and click on the Shared Match filter. The resulting list of shared matches create a new associative network of shared matches between the DNA tester (A) and the original shared match (C). I refer to the new associative list of shared matches as “D” matches. While “D” matches both “A” and “C”, the “D” match does not have to match “B” because DNA is inherited randomly. However, the “D” match should match some of the other original “C” matches for “A” and “B”. The image below visualizes these relationships.

In reviewing the list of new “D” matches, the idea is to find “D” matches who are closely related to “C” and have a family tree where you can potentially determine how “C” is then connected to either “A”, “B”, or one of the sub-clusters to which other “C” matches have been categorized. Closely related is defined as parent-child, sibling, or niece/nephew, and it is quite probable close relations do not share multiple ancestral lines. The image below exemplifies how I identified sub-cluster membership for some of the “C” matches without trees using Enhanced Shared Matches.

You can additionally use viewed match switching (see Part 2 of this series) to ensure the observed “D” match with a tree has some of the same matches that “C” and “A” also share, i.e., genetic network membership.

Enhanced Shared Matches Benefit. For matches without a family tree, the Enhanced Shared Matches filter can identify other closely related matches who have trees permitting you to associate the match with an identified sub-cluster.

It’s may not always be necessary to find “D” matches who are closely related to categorize the treeless “C” match into a sub-cluster especially if none of their close relations have tested. You may be able to use other high confidence “D” matches (1st or 2nd cousins) to allocate the treeless “C” match into a sub-cluster especially if the progenitor of the potential sub-cluster to which you intend to categorize the treeless “C” match into was born in the early 1800s.

More explicitly, the common ancestor for first or second cousins would likely be a grandchild of the progenitor who was born in the early 1800s. A grandchild relationship creates a two-generation separation from the progenitor providing greater confidence with the match categorization into the sub-cluster. Even with half-relationships or removed generations for DNA testers, the generational margin of error sufficiently covers the probability that the “C” and “D” matches are part of the same sub-cluster. According to the Shared cM Project, first to second cousins once removed share on average 866 to 122 cM. Note, these assumptions may not hold for endogamous populations where individuals share more DNA than expected.

For example, one of the sub-clusters in the Hill Genetic Network image presented earlier was for Joseph M. Keel, whose two children were born about 1802. I had a treeless “C” match who shared 132 cM with another match (D) who descended through Joseph’s son John Keel (see image below). The shared cM is at the lower bounds of acceptability (i.e., 122-866 cM), but given both are members of the same genetic network, it is quite probable both “C” and “D” descend from John Keel.

Conclusion

Using Enhanced Shared Matches, I was able to construct an overall larger genetic network of shared matches and provide greater confidence in the identified sub-clusters by having fewer “unknown” matches and greater number of matches within each sub-cluster. I now have several unlinked family clusters, i.e., Keel, Linn, Harris, and Clark, to investigate through additional DNA analysis and documentary research. I will also look for these clusters and perhaps other clusters within William Hill’s (1775-1836) other descendants whose kits I have access. These individuals provide greater coverage of William Hill’s DNA passed down to his other descendants that my cousin (A) did not inherit.

Part 6 of the Genetic Network blog series describes how to find the same genetic network across multiple DNA testing websites.

---

---

Acknowledgment: The image used within the header at the top of the blog post was created using Microsoft’s Copilot AI-powered assistant (DALL-E 3) and added to the title slide. AI tools were not used to generate the blog’s intellectual content or provide writing assistance. The post was authored solely by me.

---