Chromosome 2 Triangulation Analysis

The Donaghmoyne Network case study has long faced a structural limitation in its DNA evidence base. The Hamall direct line — Owen Hamall (1847–1898) → Thomas Henry Hamall (1880–1938) → Thomas Eugene Hamall (1904–1967) → the subject's father — narrowed in successive generations to single surviving children carrying the Hamall paternal inheritance forward. The subject's father, Thomas Kenny Hamall, died without testing his DNA. No paternal-line first, second, or third cousins exist. The five children of Thomas Kenny who have tested are siblings to one another and therefore share each other's DNA in ways that make sibling-only data structurally insufficient for paternal-line triangulation.

The implication for the Donaghmoyne research has been significant. Each of the four documented Donaghmoyne couples has been corroborated within its own descendant lines: the James Hamill & Ann Gartlan sibling group through Peter Hamill's 1949 death certificate and supporting DNA, the Owen Hammel & Ann King family through Wisconsin descendants, the Susan Hamill & Charles McCanna family through Joliet descendants. But the cross-line evidence — whether Henry Hamall, Owen Hammel, James Hamill of Dian, and Susan Hamill McCanna's father were brothers, first cousins, or more distantly related — has been carried by autosomal matches at threshold levels (typically 8–34 cM) that could be explained by either true descent from a common ancestor or by chance.

The December 2025 DNA Evidence Analysis page concluded that segment triangulation and Y-DNA testing would be the next steps required to strengthen the case beyond suggestive matching. This page reports the results of the segment-triangulation work that has since been completed, along with a methodological development that resolves the Chicago-line bottleneck without requiring Y-DNA: the construction and validation of a reconstructed paternal kit representing Thomas Kenny Hamall's paternal genome, built from his children's DNA using his wife's kit as a phasing reference.

The reconstructed kit makes possible the analytical step that was previously unavailable to this case study: each cluster match can now be tested directly against the subject's father's reconstructed paternal DNA, producing individually documented paternal-inheritance confirmation in addition to the cluster-level patterns previously available. The combination of mathematical triangulation across multiple platforms with reconstructed-kit one-to-one paternal verification produces segment evidence at the level of rigor sought by the December framework — and makes claims now possible that the earlier evidence base could not support.

The reconstructed paternal kit was produced through the Borland Genetics third-party reconstruction workflow, which builds a parent's genome from the DNA of multiple children and a co-parent. The conceptual basis is straightforward. Each child inherits exactly half of their DNA from each parent, and the inheritance patterns of full siblings overlap on roughly 50% of their genomes. By comparing each child's DNA against their mother's kit and identifying the segments that did not come from her, the algorithm extracts each child's paternal contribution. The four resulting paternal-only datasets are then merged into a single composite kit representing the deceased father's paternal genome.

For this analysis, the inputs were four children's kits — three 23andMe kits (the subject and two siblings, C. and K.) and one FTDNA kit (A., who was unable to provide saliva samples and therefore tested via FTDNA's cheek-swab protocol) — plus the maternal kit (the subject's mother, Barbara O'Brien Hamall). The mother's kit functions as a phasing reference: any DNA in a child's kit that matches the mother's kit is excluded from the paternal extraction, leaving only the paternal contribution. The merged paternal dataset was uploaded to GEDmatch as a research kit, where it can be compared one-to-one against any other kit in the GEDmatch database.

One operational constraint of the resulting kit is significant for the methodology that follows. GEDmatch designates Borland-derived reconstructed kits as research kits, which restricts them to one-to-one comparisons. The standard one-to-many search that allows a tester to discover unknown matches is not available for research kits. Identifying candidate paternal matches therefore proceeds through indirect methods — running one-to-many searches on the subject's own kit and her siblings' kits, identifying matches that appear paternal-side based on shared-match patterns, and then running each candidate one-to-one against the reconstructed kit for confirmation.

The reconstructed kit's accuracy was validated by comparing it one-to-one against each of Thomas Kenny Hamall's five children. For a true parent-child comparison, GEDmatch's one-to-one tool returns characteristic statistics: a total Half-Identical Region (HIR) shared cM value of approximately 3,400–3,600 (each child shares roughly half a parent's genome), an estimated MRCA of one generation, near-complete coverage of the comparable SNPs, and a high full-identical-by-descent percentage indicating both alleles match on most compared positions. Any reconstruction that fails to produce these characteristics for known children is producing artifactual rather than accurate inheritance data.

Four of the five comparisons were against children whose kits contributed to the reconstruction itself — three from 23andMe and one from FTDNA. These comparisons confirm the algorithm successfully integrated each input — but because each contributing sibling's data is part of the reconstruction, their comparisons reflect both true paternal inheritance and reconstruction-input overlap. They are not, strictly speaking, independent tests of reconstruction accuracy.

The fifth comparison was against E., the subject's sister who tested at Ancestry rather than 23andMe and was not used in the reconstruction. Her kit was uploaded to GEDmatch from her Ancestry raw data file. Her comparison against the reconstructed kit is therefore structurally independent: her DNA is not part of the reconstruction's input, so any match between her kit and the reconstructed kit reflects genuine inheritance from the same biological father — not algorithmic overlap. This independent platform validation is the gold-standard test for reconstruction accuracy.

The reconstruction methodology therefore drew on three independent SNP-chip platforms across input and validation: 23andMe (three contributing siblings), FTDNA (one contributing sibling, A.), and Ancestry (E. as independent validation reference). Cross-platform input diversity strengthens reconstruction integrity by reducing dependence on any single platform's chip design.

All five comparisons return the GEDmatch one-to-one tool's estimate of one generation to MRCA — confirming the reconstructed kit functions correctly as a parent-child comparator. The four contributing siblings show HIR coverage between 85.90% and 99.92%, with the variation reflecting each sibling's individual role in the reconstruction algorithm. A.'s near-perfect 99.92% coverage suggests her kit served as a primary phasing anchor; the others' coverage reflects fill-in contributions across the algorithm's processing.

E.'s comparison is the most analytically important result in the table. Her Ancestry kit was tested on a different SNP chip than the 23andMe kits used in the reconstruction, and the comparable SNPs between the two chips are limited — only 148,461 SNPs were available for the comparison versus 423,000–436,000 for the contributing siblings. Despite this reduced SNP density, E.'s comparison returned a Total HIR of 2,500.8 cM, a 1-generation MRCA estimate, and a full-IBD percentage of 67.14%. These values are consistent with a true parent-child relationship measured at lower SNP resolution. With the same SNP count available to her as her siblings had, her HIR would almost certainly land in the 90–95% range. The proportional pattern at the available SNPs already confirms the reconstruction produces accurate paternal inheritance data.

The reconstructed kit could be improved in coverage by adding E.'s Ancestry kit as a fifth sibling input. Five-sibling reconstructions typically reach 90–95% paternal coverage versus the 75–90% range of four-sibling reconstructions. Adding E. would close some of the segment gaps, potentially surfacing additional paternal matches that the current reconstruction does not detect.

However, adding E. to the reconstruction would eliminate her function as the independent platform validation reference. Any subsequent comparison against the reconstructed kit would no longer have a structurally independent test point. For a research project that aspires to BCG portfolio standards and that may be evaluated by reviewers familiar with reconstruction methodology, the independent validation reference has substantial methodological value.

The decision documented for this analysis: E.'s kit is retained as the validation reference. She has not been added to the reconstruction. Should subsequent research identify a specific paternal coverage gap that her DNA could fill, this decision is reversible, but the marginal coverage improvement does not currently justify trading away the independent validation. This decision is recorded in the project's research log.

Before any segment was painted or any triangulation claim was advanced, the chromosome 2 cluster was identified through four independent paths across three DNA-testing platforms. Cross-platform convergence on the same chromosomal region constitutes robust cluster identification: no single platform's algorithm or matching threshold determined the finding.

The first identification path used 23andMe's Cluster Relatives in Common feature with P.H. (the highest-cM cluster member at 38.5 cM with the subject) as the anchor. Selecting P.H. and requesting in-common-with matches surfaced a coherent grouping of paternal-side matches: S.M., M.R., B.R., T.S., D.O., J.L. and others. The 23andMe cluster did not name specific chromosomes, but the matches' segment data (visible on each match's individual page) consistently included shared chromosome 2 regions in the 35–60 megabase area.

The second identification path used GEDmatch's Compact Segment Mapper applied to the subject's top 15 matches. The visualization rendered each match's largest shared segments across all chromosomes simultaneously, and chromosome 2 stood out immediately as the densest stacking area. Eight to nine of the top fifteen matches had segments overlapping in roughly the same chromosome 2 region, with no comparable density on any other chromosome.

The third identification path used GEDmatch's "people who match both kits, or 1 of 2 kits" search applied to the subject's kit and P.H.'s kit. This surfaced individuals matching both — a list that overlapped substantially with the 23andMe cluster but added several GEDmatch-only matches (J.S., A.D., D.O.) not visible through 23andMe's tools.

The fourth identification path used MyHeritage's chromosome browser, which independently surfaced a chromosome 2 paternal cluster among the subject's MyHeritage matches. The MyHeritage cluster overlapped substantially with the GEDmatch and 23andMe clusters but added several MyHeritage-exclusive matches (A.W., A.C., A.B., R.S.) — particularly testers who had uploaded to MyHeritage from Ancestry and were therefore not visible through 23andMe.

The four identification paths converged on the same chromosomal region from different starting points and different match pools. This convergence is the strongest available pre-triangulation signal that the cluster represents genuine shared ancestry rather than algorithmic artifact.

MyHeritage's chromosome browser includes a triangulation tool that mathematically confirms when multiple selected matches share the same DNA segment with each other and with the kit owner simultaneously. Unlike one-to-one comparison, which only verifies bilateral sharing, triangulation establishes that all selected matches share the same physical DNA segment — a pattern explainable only by descent from a common ancestor who passed that segment forward through multiple lines.

Two triangulation groups were established for the chromosome 2 cluster, with overlapping but not identical membership. The two groups differ because MyHeritage's algorithm computes the minimum shared region across whatever subset of matches is selected. Different subsets produce different minimum-region boundaries, but both groups represent the same underlying ancestral segment.

TG2's coordinates are nested within TG1's — TG2 (44,599,253 – 50,854,610) sits inside TG1 (44,343,676 – 51,515,672). This is the expected behavior when MyHeritage's algorithm computes the minimum shared region across different match subsets. TG2 represents the conservative core: the smallest segment that all seven of its members definitively share. TG1 represents the broader region shared by its six members. Both groups identify the same underlying ancestral segment.

Combining the membership of both groups yields nine unique MyHeritage matches confirmed mathematically triangulating on chromosome 2: B.H., A.W., J.K., R.S., M.G., B.R., A.C., A.B., R.C. Four matches (B.H., A.W., M.G., J.K.) appear in both groups, providing the strongest individual-match triangulation confirmation; the remaining matches appear in one group but not the other due to where their individual segments end relative to the algorithm's minimum-region calculation.

Mathematical triangulation establishes that multiple matches share the same chromosomal segment, but it does not establish whether that shared segment was inherited through the paternal or maternal line. For the Donaghmoyne research question, the distinction is essential: a segment shared through the maternal O'Brien-Robertson line would not bear on the Hamall paternal-line hypothesis at all.

The reconstructed paternal kit allows direct paternal-line confirmation. For any match in the chromosome 2 cluster, a one-to-one comparison against the reconstructed kit returns a definitive result: a confirmed paternal segment, no significant shared cM (indicating a maternal-line connection), or a coverage gap (the shared region falls in a part of the paternal genome not captured by the reconstruction).

One operational note constrains the workflow. GEDmatch is currently the only platform supporting one-to-one comparisons against research kits. The reconstructed kit cannot be tested against MyHeritage-only matches (A.W., A.C., A.B., R.S., R.C.) because those matches' raw DNA has not been uploaded to GEDmatch. Their triangulation through MyHeritage's mathematical tool stands as confirmed shared segment evidence, but their paternal inheritance cannot be independently verified through the reconstructed kit. For the matches available on GEDmatch, the one-to-one results below constitute the most rigorous form of paternal-inheritance documentation currently available for this case study.

P.H.'s comparison against the reconstructed kit returned three confirmed paternal segments — the chromosome 2 main cluster region, plus two additional segments on chromosome 4 (7.2 cM) and chromosome 16 (7.3 cM). The two additional segments are above the standard 7 cM IBD reliability threshold but at its lower bound, and they are flagged accordingly in the project's research log.

Multi-chromosome paternal sharing is significant for two reasons. First, it indicates that P.H.'s total shared cM with Thomas Kenny Hamall (across all three segments) is meaningfully higher than P.H.'s 38.5 cM total shared cM with the subject. The combined evidence is consistent with a 3rd-cousin to 3C1R relationship range — closer than the typical 4th-cousin range that single-cluster-segment matches tend to occupy. Second, multi-chromosome sharing supports a closer MRCA estimate. A single shared segment can survive recombination across many generations; multiple independent segments are more likely to indicate fewer generational distances. The combined evidence places P.H.'s most recent common ancestor with the subject's father in the 1820s–1850s Donaghmoyne generations, consistent with the documented timeframe of the four-couple Donaghmoyne network.

Beyond the segment-level evidence on chromosome 2, the cluster's matches share a distinct surname signature in their documented ancestry. Cross-tabulating the documented family connections of all cluster matches yields a recurring set of surnames concentrated geographically in southeast County Monaghan and adjacent County Louth. The surname pattern is independent corroboration of the segment evidence: it is not derived from the DNA itself but from primary-source documentation of each match's ancestral lineage, and it concentrates in a specific micro-region that would not be expected if the cluster represented broader Irish or general European ancestry.

The geographic concentration is the central analytical fact. The surnames distribute primarily across four parishes — Donaghmoyne, Magheracloon, Inniskeen, and Carrickmacross — in southeast County Monaghan, with extensions into adjacent County Louth (Termonfeckin and Dundalk). This corresponds to a coherent pre-Famine Catholic community of approximately 15–25 square miles, within which the documented marriage patterns show repeated intermarriage across the same surname families over three generations.

The pattern is genetically and demographically consistent with an endogamous parish community: a Catholic population restricted in marriage choice by religious affiliation, geographic isolation, and limited transportation, in which the same families intermarried repeatedly. For DNA matching, endogamy produces exactly the signature observed here — multiple independent surname lines sharing the same chromosomal segments because they all descend from a small founding population that contributed DNA broadly across the community.

What the surname signature does not establish is any specific most-recent-common-ancestor relationship between the subject's direct line and any individual cluster member. Identifying that a cluster of matches descends from a small Donaghmoyne population is a population-level finding, not a genealogical one. The genealogical work — identifying the specific ancestral couple through whom the subject and a given match share descent — requires documentary research in pre-Famine and early-Famine Catholic parish records.

The full analytical workflow is most clearly demonstrated through specific worked examples. Each of the three below shows how the methodology was applied to one match, what the evidence established, and what remains open for future research.

The December 2025 DNA Evidence Analysis page used a four-tier evidence framework: PROVEN, SUGGESTIVE, EXPLORING, and INDIRECT. The new chromosome 2 evidence requires adding a higher tier and updating where specific connections sit. The framework below carries forward the December tiers and adds a new top-level CONFIRMED tier for findings now backed by segment-level triangulation plus reconstructed-kit paternal verification plus multi-platform cross-validation.

An endogamous Donaghmoyne genetic population exists as an identifiable cluster, demonstrated through cross-platform segment evidence at three layered levels: independent identification through multiple search paths, mathematical triangulation across nine MyHeritage matches, and direct paternal-inheritance confirmation through one-to-one comparison against the subject's father's reconstructed kit for at least nine GEDmatch-accessible cluster matches.

The subject's paternal line descends from this population. The reconstructed paternal kit demonstrates that segments shared between cluster matches and the subject also are shared between cluster matches and the subject's father directly, eliminating maternal-line ambiguity for the chromosome 2 cluster region.

Cross-line genetic continuity is documented across at least 13 independent matches and 17 recurring surnames concentrated in southeast Monaghan and adjacent Louth — a geographic specificity inconsistent with general Irish or European ancestral matching.

The Chicago paternal-line bottleneck that previously limited the case study's evidence base no longer constrains paternal-line analysis. The reconstructed paternal kit provides a stable comparator for any future paternal candidate match transferred to GEDmatch.

Specific most-recent-common-ancestor relationships for any pair of cross-line matches at five-or-more generations distance are not established by this analysis. Identifying that the subject's paternal line shares an endogamous Donaghmoyne ancestor with a given cluster match is a population-level finding, not a genealogical one. The documentary work that would identify the specific ancestral couple — for example, whether Henry Hamall, Owen Hammel, James Hamill of Dian, and Susan Hamill McCanna's father were brothers, first cousins, or more distantly related — remains to be completed.

Pre-1830 Donaghmoyne genealogy has not been resolved by this analysis. Catholic parish baptismal and marriage registers for Donaghmoyne begin in 1834. Earlier ancestral identification requires alternative pre-Famine sources (Catholic Qualification Rolls 1778–1790, surviving estate records, the Morpeth Roll 1841, Tithe Applotment Books 1823–1838) that have not yet been systematically searched for the relevant family lines.

User-tree attributions of pre-1800 ancestry remain suggestive but unverified. The recurring James Hamell I (1652–) ancestor in multiple cluster trees is an example: the attribution may reflect community-shared use of an early-published genealogy rather than independent primary-source documentation. Until that source is identified and evaluated, the attribution is treated as a research lead rather than as evidence supporting any specific finding.