Mapping a Fresh Chromosome 18

Introduction

Over the past several months, I have spent considerable time reviewing the Ancestry DNA matches of my great-uncle, Jeffrey James Borland (1943–2016), whom we called Jeff. Much of that work has involved identifying clusters of related matches, tracing their genealogical connections, and determining where they belong within Jeff’s extensive family tree.

At the same time, I have been working on another, closely related project: identifying Jeff’s paternal cousins and mapping the chromosome segments they share with him. These relationships are particularly valuable for Borland Genetics, where I use DNA from living relatives to reconstruct portions of the genomes of our deceased ancestors. By identifying which segments Jeff inherited from particular ancestors, I can create reconstruction masks that help assign DNA to the appropriate ancestral genomes.

Jeff’s father, Weldon Earl Borland (1906–2002), was my direct paternal great-grandfather. Consequently, Jeff’s paternal DNA provides an especially valuable window into the ancestors we share through Weldon and his family. One of Jeff’s daughters has also taken an autosomal DNA test, giving me another close relative whose inherited segments can help clarify the transmission of DNA across generations.

My initial approach to mapping Jeff’s chromosomes was relatively straightforward. I examined his DNA matches at MyHeritage and looked for people I recognized, either because I knew them personally or because I had already identified them through my research into Jeff’s Ancestry DNA matches. When I recognized a match and knew how that person was related to Jeff, I could examine the chromosome segments they shared and begin assigning those regions to known ancestral lines.

This approach proved surprisingly productive. I was able to identify useful matches on every one of Jeff’s autosomes except one: chromosome 18. Despite the progress elsewhere in his genome, chromosome 18 remained a conspicuous blank in my mapping work. In the Borland Genetics HIR Mapper, it appeared as an unmapped black rectangle, with no ancestral assignments to distinguish one portion from another.

There is, however, one important piece of information already available. I share a segment of approximately 35 centimorgans (cM) with both Jeff and his tested daughter, beginning at the western tip of chromosome 18. That establishes a recognizable shared region—a segment we can locate on the chromosome—even though it does not yet identify the more distant ancestor from whom the segment originated.

Rather than continue searching through matches by eye, I approached chromosome 18 systematically, using the complementary capabilities of GEDmatch, MyHeritage, DNA Painter, and Ancestry—chromosome segment comparisons, genealogical research, and shared-match analysis together.

This article documents that work: the methods I used, the evidence I found, the places where the data ran out, and the reasoning behind the assignments I could defend. On Jeff’s paternal copy of chromosome 18, I separated maternal from paternal segments, grouped Weldon-side matches into four regional clusters, and assigned two of those regions—to James Couthren Borland through vertical parity and to Miller and Wheeler ancestry through conventional research on the Nancy cluster. The western Sarah cluster and the eastern Christina cluster remain unresolved.

The ancestral assignments I could support can be incorporated into Borland Genetics using its Extract Segments tool, so the newly identified DNA can contribute to reconstructing our common ancestors. The goal was never merely to erase the black rectangle in one pass, but to show how an apparently uninformative chromosome can be moved toward a defensible partial map—and how to recognize when the evidence is not yet strong enough to go further.

Applying the Reverse Phase Technique

The first logical step in mapping Jeff’s chromosome 18 was to separate his paternal DNA, inherited from Weldon Earl Borland, from his maternal DNA, inherited from Vivian. Ancestry provides a useful starting point through its SideView technology, which automatically groups many DNA matches by parental side. Neither MyHeritage nor GEDmatch offers an equivalent automatic assignment of matches to paternal and maternal ancestry, however, so I needed another way to distinguish the two sides of Jeff’s genome.

I decided to begin with GEDmatch because it provides phasing tools that make this task relatively straightforward. Although neither of Jeff’s parents underwent DNA testing, one of his daughters tested with 23andMe. This made it possible to apply a technique I call reverse phasing, using the DNA of a child to separate the two chromosome copies of a tested parent.

Ordinarily, phasing involves using a parent’s DNA to determine which portions of a child’s genome came from each side of the family. Reverse phasing approaches the problem from the opposite direction. By comparing Jeff’s DNA with that of his daughter, GEDmatch can identify the alleles he transmitted to her and distinguish them from the alleles he did not transmit.

Using GEDmatch’s Phasing and Evil Twin tools, I created two derivative kits. The first, which I will call the in-phase kit, represents the DNA Jeff passed to his daughter. The second, his paternal evil twin kit, represents the complementary DNA from the opposite copies of his chromosomes, consisting of the alleles he did not transmit to her. Both kits can be compared against other GEDmatch participants, generating DNA matches that can be examined using the site’s chromosome comparison tools.

An important point is that these kits are not inherently paternal or maternal in the genealogical sense. Each represents one of the two complementary sets of DNA from Jeff’s genome. Because recombination can occur, the transmitted chromosome may contain stretches inherited from both of that parent’s parents. The corresponding evil twin chromosome contains the complementary stretches.

For example, wherever the in-phase kit contains DNA Jeff inherited from Weldon, the evil twin kit should contain the corresponding DNA inherited from Vivian. If a crossover occurred during Jeff’s transmission to his daughter, the parental origins represented by the two kits would switch at that location. The in-phase kit might begin with Weldon’s DNA and continue with Vivian’s DNA after the crossover, while the evil twin would show the opposite arrangement.

The task, therefore, was to determine the parental origin of representative matches along chromosome 18 and look for any changes in that assignment. By comparing the starting and ending positions of matches in the two kits, and identifying those matches genealogically, I could establish which portions of the chromosome represented Weldon’s ancestry and which represented Vivian’s. A change from paternal to maternal matches within the same kit would provide evidence of a recombination boundary.

Initially, this was not as simple as recognizing familiar names. Apart from my own match, I did not recognize the chromosome 18 matches in either GEDmatch kit by inspection. However, I was able to search for these individuals among Jeff’s Ancestry DNA matches and could use SideView assignments and known family relationships to determine their ancestral origins.

Those searches provided enough identifiable matches to establish a consistent pattern across chromosome 18. Every informative match associated with the in-phase kit belonged to Jeff’s paternal ancestry through Weldon Earl Borland. Conversely, the informative matches associated with the evil twin kit belonged to his maternal ancestry through Vivian.

No switch between paternal and maternal ancestry was detected anywhere along chromosome 18. Had a recombination occurred within the regions covered by the informative matches, I would have expected to encounter a change in the parental assignments as I moved along the chromosome. Instead, the assignments remained consistent from one end to the other.

The evidence therefore supports the conclusion that Jeff transmitted his paternal chromosome 18 to his daughter without an observable crossover, while the complementary evil twin kit represents his maternal chromosome 18. Although a crossover in a region lacking informative matches cannot be absolutely excluded, no such event was indicated by the available data.

Segment Triangulation Using MyHeritage

Having established which of Jeff’s chromosome 18 copies came from his father, Weldon Earl Borland, and which came from his mother, Vivian, the next step was to apply that information to his MyHeritage DNA matches. Unlike GEDmatch, MyHeritage does not provide the phasing tools I had used to distinguish the two parental chromosomes. However, it does provide chromosome comparisons and triangulation capabilities that make it possible to determine which side of Jeff’s ancestry an individual matching segment belongs to.

Before MyHeritage discontinued its mass DNA match export feature, I had exported Jeff’s matching segments to DNA Painter. This proved particularly useful because I could now view the positions of his chromosome 18 matches together, rather than examining them individually. Starting at the western end of the chromosome and working eastward, I could identify overlapping segments and systematically investigate their parental origins.

The underlying principle is straightforward. If an unidentified match shares a particular chromosome segment with Jeff, and that individual also matches a known paternal relative of Jeff on the same segment, with all three sharing the same DNA, the segment can be assigned to Jeff’s paternal chromosome. Conversely, triangulation with a known maternal relative identifies a segment belonging to his maternal chromosome. Simply having overlapping matches is not sufficient; the matches must also share the segment with one another.

The reverse-phased GEDmatch kits provided the reference information needed to carry out this process. My task was to identify representative GEDmatch matches from different regions of chromosome 18 who had also tested at MyHeritage. These individuals could then serve as reference matches for triangulating other MyHeritage segments whose parental origins were still unknown.

The first approximately 35 cM at the western end of chromosome 18 was the easiest region to investigate because I already knew that I shared this segment with both Jeff and his tested daughter. Since my relationship to Jeff is through his father, Weldon, I could use my own DNA as a paternal reference. MyHeritage matches who triangulated with Jeff and me across this region could therefore be assigned to Weldon’s side of the family.

Moving eastward, I continued the same process using other reference matches whose paternal or maternal assignments I had already established through GEDmatch. For each unidentified MyHeritage segment, I examined its overlap with the reference segments and used triangulation to determine whether it belonged to Jeff’s paternal or maternal chromosome. Once I established the assignment, I updated the segment’s parental designation in DNA Painter.

This was a methodical process, but it was also productive. Over the course of the next 48 hours, I was able to classify 146 of Jeff's paternal chromosome 18 matches, assigning 92 to his paternal side and 54 to his mother Vivian's side. These included every chromosome 18 match sharing at least 20 cM with Jeff at MyHeritage, along with numerous smaller matches.

The important result was that chromosome 18 was no longer an undifferentiated collection of unidentified matches. By combining GEDmatch reverse phasing, MyHeritage triangulation, and DNA Painter’s chromosome mapping interface, I had assigned parental origins to a substantial number of segments without having DNA tests from either of Jeff’s parents.

For the purposes of this investigation, the 92 matches assigned to Weldon’s side were particularly important. They provided a starting collection of chromosome 18 segments that could now be examined genealogically to determine which of Weldon’s ancestral families they represented. The next challenge would be to move beyond identifying the correct parent and begin assigning those segments to specific ancestors.

Resisting Temptation

I must confess that, despite my determination to approach chromosome 18 systematically, I couldn’t resist taking a peek at Ancestry’s ethnicity chromosome painter to see what it had to say about the chromosome. Its verdict was remarkably simple: the entire chromosome was attributed to “Ulster, Ireland.”

Now, that was tempting!

Weldon’s father, James Couthren Borland (1877–1943), known to the family as “J.C.,” had ancestry firmly rooted in Ulster. The Borlands emigrated to Pennsylvania from Kilraghts Parish in County Antrim, within the historic province of Ulster. By contrast, I am not aware of any ancestors of Weldon’s mother, Lizzie Alberta Miller (1875–1975), who came specifically from Ulster.

Wouldn’t it be convenient to declare chromosome 18 a Borland chromosome and move on?

Unfortunately, genetic genealogy rarely rewards such shortcuts. Many of Lizzie’s ancestors came from other parts of the British Isles, and I have considerably less confidence in ethnicity prediction algorithms when they attempt to distinguish populations with closely related genetic histories. With seemingly every ethnicity update, we hear from test-takers whose Scottish ancestry has become Irish, or whose Irish ancestry has suddenly turned Scottish.

These estimates can provide useful clues, particularly when the populations involved are genetically and geographically quite distinct. But distinguishing between neighboring populations in the British Isles is another matter entirely. An ethnicity assignment of “Ulster, Ireland” does not establish that a chromosome segment came from a particular ancestor, much less prove that it belongs to the Borland rather than the Miller side of Weldon’s pedigree.

So, tempting as it was, I resisted assigning chromosome 18 to J.C. Borland on the strength of the ethnicity painter alone. The prediction was worth keeping in mind as a possible clue, but the actual ancestral assignments would have to come from triangulated DNA segments and genealogically established relationships.

Back to the evidence!

Surveying the Paternal Landscape

With Jeff’s chromosome 18 matches separated into paternal and maternal groups, I could finally concentrate on the matches inherited through his father, Weldon Earl Borland. When I examined their positions in DNA Painter, an encouraging pattern emerged: the identified paternal matches fell quite neatly into four distinct spans of the chromosome.

For convenience, I decided to treat each span as a separate cluster, naming it after a representative match whom I designated the lead donor. Generally, I selected the individual sharing the longest segment with Jeff in that region, excluding my brother and myself. These names are simply organizational labels for the investigation. At this stage, I have not established that all the members of a cluster descend from the same ancestral couple, only that their matching segments occupy the same general region of Jeff’s paternal chromosome 18.

Starting at the western tip of the chromosome, where my brother and I also match Jeff, I identified the Sarah cluster. Sarah shares a 24 cM segment with Jeff and serves as the lead donor for a group consisting of Sarah, Robert, Joan, as well as 8 matches at below 20 cM.

Moving eastward, the next group is the Fred cluster, led by Fred’s 32 cM segment. While Fred's match is significant, the other 16 matches I identified on this span were all less than 20 cM.

Farther east is the Nancy cluster, an extensive cluster of 63 matches, with Nancy sharing 28 cM with Jeff. There are numerous 20+ cM matches on this span of chromosome.

Finally, at the eastern end of chromosome 18, I identified the Christina cluster, representing approximately the last 10 to 15 cM of the chromosome. This region presented a different challenge because I had no sizable paternal matches to work with. I selected Christina, whose 9 cM segment appeared to be one of the more reliable matches available, to serve as the lead donor for this portion of the chromosome.

The four working groups can be summarized as follows:

Conceptually, the next step is simple enough. I need to examine the family trees of the matches in each cluster, identify ancestors they may share with Weldon, and determine whether those connections can explain the corresponding chromosome segments. Ideally, this will allow me to assign each of the four regions to an ancestor, or ancestral couple, in Weldon’s family tree.

In practice, of course, things are rarely quite that cooperative.

Some matches have useful family trees, but many do not. For most of these individuals, I expect to have to build substantial portions of their pedigrees myself, working backward through census records, vital records, and other genealogical sources until I find a plausible connection. For the smaller matches, particularly those sharing fewer than 10 cM, the common ancestor may be so far back that reconstructing the necessary generations will prove difficult or even impossible.

Still, the chromosome that began as an uninterrupted black rectangle has now become four smaller research problems, each with a set of candidate matches and a defined region to investigate.

Four clusters, four genealogical puzzles. Time to start building some trees!

A Breakthrough in the Nancy Cluster

Of the four groups I had identified on Jeff's paternal chromosome 18, the Nancy cluster initially looked especially promising. With 63 matches, including several sharing more than 20 cM with Jeff, it offered a substantial collection of potential relatives to investigate.

As it turned out, the breakthrough came from Nancy herself.

I began building out Nancy's family tree, tracing her ancestral lines backward in search of a connection to Weldon Earl Borland. Eventually, I discovered that Nancy was related to Weldon's maternal grandfather, Samuel Hoobler Miller (1851–1912), through the family of Samuel's mother, Mary Ann (Wheeler) Miller (1822–1858).

What made the discovery particularly interesting was that Nancy's connection to Mary Ann Wheeler was not limited to a single ancestral line. Instead, I found several lines of relationship, all leading back to families associated with St. Paul's Parish in Baltimore, Maryland.

This was familiar territory. Mary Ann Wheeler and her extended family had already figured prominently in my earlier investigation, Who Is John Gaut?, in which I examined a cluster of DNA matches associated with the Wheeler family of colonial Maryland. Now, an entirely separate chromosome investigation had brought me back to the same family.

Given Nancy's genealogical proximity to this family, her multiple connections through the St. Paul's Parish community, and the chromosome segment she shares with Jeff, I was sufficiently confident to assign the Nancy cluster's region to Weldon's maternal ancestry, specifically through Lizzie Alberta (Miller) Borland, Samuel Hoobler Miller, and Mary Ann (Wheeler) Miller.

Vertical Parity: What We Learn from an Independent Reconstruction of Weldon

The Fred cluster presented a different sort of challenge. Although Fred shares a substantial 32 cM segment with Jeff, he has no family tree or useful identifying information that would allow me to reconstruct his ancestry. Ordinarily, this might bring the investigation to a temporary halt. Without knowing who Fred is or how he might be related to Jeff, how could I determine which of Weldon's ancestors contributed this portion of chromosome 18?

Fortunately, there is another way to approach the problem, using a principle of visual phasing known as the vertical parity rule.

The principle is straightforward. Every person carries two copies of each autosome, one inherited from each parent. When a parent has two children, each child receives a recombined copy of that parent's chromosomes. At any particular position, the children may have inherited DNA from the same one of their parent's two chromosome copies, or they may have inherited DNA from opposite copies.

If two children inherited opposite copies at a particular position, one received DNA that their parent had inherited from their paternal side, while the other received DNA from their maternal side. By establishing the ancestral origin of the DNA transmitted to one child, we can therefore infer the origin of the corresponding DNA transmitted to the other.

This is particularly useful in my family because Weldon Earl Borland had two sons whose descendants have undergone DNA testing: Jeffrey James Borland, whose chromosome 18 I am attempting to map, and John Earl Borland, Jeff's paternal half-brother and my grandfather.

My brother Steven and I have both tested, and we inherited paternal DNA across the region occupied by the Fred cluster through our grandfather John Earl Borland. This gives us a second line of evidence for reconstructing Weldon's chromosomes, independent of the matches identified in Jeff's GEDmatch and MyHeritage kits.

In an earlier investigation, documented in my Bright Cluster article, I established that Steven and I inherited a substantial portion of this region from Harrison Miller, the father of Samuel Hoobler Miller. The relevant line of inheritance runs from Harrison Miller through Samuel Hoobler Miller, Lizzie Alberta Miller, Weldon Earl Borland, and John Earl Borland to our branch of the family.

That previous assignment now becomes extremely valuable.

Across the Fred cluster's span of chromosome 18, the DNA inherited through John Earl Borland represents Weldon's maternal ancestry, specifically the Miller family. Jeff, however, does not match the corresponding DNA inherited through John Earl's line.

Applying the vertical parity rule, this indicates that Weldon transmitted opposite copies of his chromosome 18 to his two sons in this region. John Earl received the DNA Weldon had inherited through his mother, Lizzie Alberta Miller, while Jeff received the corresponding DNA Weldon had inherited through his father, James Couthren Borland (J.C.).

In other words, I could assign the Fred cluster to J.C. Borland without ever identifying Fred's genealogical connection to the family.

There was another particularly satisfying aspect of this discovery.

Immediately adjacent to the Fred cluster is the Nancy cluster, which I had just assigned to Lizzie Alberta Miller, her father Samuel Hoobler Miller, and his mother Mary Ann Wheeler. At first glance, the two neighboring regions of Jeff's chromosome 18 appear to have different ancestral origins: the Fred region from J.C. Borland and the Nancy region from the Miller family.

But when we consider the two chromosome copies Weldon inherited, the arrangement makes sense.

The DNA transmitted to Jeff across the Fred region came from Weldon's paternal chromosome, while the DNA transmitted to John Earl across that same region came from Weldon's maternal chromosome. The latter had already been assigned through Samuel Hoobler Miller to Harrison Miller.

Immediately to the east, in the Nancy region, Jeff's DNA is also attributable to Samuel Hoobler Miller, this time through Samuel's mother, Mary Ann Wheeler.

Thus, when we reconstruct the two copies of Weldon's chromosome 18 rather than looking only at the DNA he transmitted to Jeff, the evidence is consistent with a continuous stretch of DNA inherited from Samuel Hoobler Miller along Weldon's maternal chromosome. Across the Fred region, that DNA was passed to John Earl; across the neighboring Nancy region, DNA from Samuel's maternal side was passed to Jeff.

This provides a useful consistency check on the Nancy cluster assignment. The evidence from the Bright Cluster investigation and the new evidence from Nancy fit together into a coherent picture of Weldon's maternal chromosome, even though the two sons inherited different portions of it. The Fred cluster had been assigned without solving Fred's family tree, and the neighboring Nancy cluster had gained an additional layer of support.

Conclusion: Knowing When to Stop

With the Fred and Nancy clusters assigned, two substantial portions of Jeff's paternal chromosome 18 had finally acquired identifiable ancestral origins. The Nancy cluster led to Mary Ann (Wheeler) Miller through conventional genealogical research, while the Fred cluster was assigned to J.C. Borland through vertical parity and an independent reconstruction of Weldon's DNA.

That left two unresolved regions: the Sarah cluster at the western end of the chromosome and the Christina cluster at the eastern end.

My next logical step was to investigate the Sarah cluster using another principle of visual phasing, known as the horizontal parity rule.

The rule states that when two siblings begin or cease matching one another at a particular point on a chromosome, the change ordinarily reflects a recombination in the DNA transmitted from their shared parent to one of the siblings. By examining the positions of these changes alongside other known chromosome segments, it may be possible to determine which sibling received the recombined chromosome.

In this case, the boundary between the Sarah and Fred regions was particularly interesting. Jeff and his paternal half-brother John Earl Borland shared DNA across the western portion of chromosome 18, as demonstrated by the segments my brother Steven and I inherited through John Earl that also match Jeff. However, that matching ended as we approached the Fred region.

One possibility was that Weldon had transmitted a recombined chromosome to Jeff, switching from one of his parental chromosome copies to the other near this boundary. If Jeff's matches with the Fred cluster began precisely where his matches with Steven and me ended, the coincidence would provide a strong clue that the crossover occurred in the transmission from Weldon to Jeff, rather than in the transmission from Weldon to John Earl.

Unfortunately, the evidence did not cooperate quite so neatly.

There is a significant gap between the end of the shared segment with Steven and me and the beginning of Jeff's match with Fred. Without a clearly established boundary between the two regions, I cannot confidently determine which of Weldon's sons received the recombined chromosome at that location.

The horizontal parity analysis therefore remains inconclusive, and although additional segment comparisons and genealogical research may eventually resolve the Sarah cluster, the available evidence does not yet justify an ancestral assignment; the Christina cluster, represented by smaller matches near the eastern end of the chromosome, likewise requires further investigation (I only identified a single small-segment match across her span), so for now I am going to leave both ends of Jeff's paternal chromosome 18 unresolved.

That does not mean the investigation was unsuccessful. Quite the contrary. Starting with a chromosome that appeared as an uninterrupted black rectangle in the Borland Genetics HIR Mapper, I used reverse phasing at GEDmatch to distinguish Jeff's paternal and maternal chromosomes, triangulation at MyHeritage to classify his matching segments, and DNA Painter to organize those segments into workable clusters. Genealogical research then identified the ancestry of the Nancy cluster, while an independent reconstruction of Weldon's chromosome provided the evidence needed to assign the Fred cluster.

The result is a substantially improved understanding of Jeff's chromosome 18 and, more importantly, a demonstration of how several complementary genetic genealogy techniques can solve problems that none of them could resolve independently. The remaining Sarah and Christina clusters will have to wait for another day; there are plenty of other promising matches in Jeff's DNA results, and for the moment, I intend to pursue some of that lower-hanging fruit.