Why DNA Tests Struggle With North American Native Ancestry — and How They Still Found Ours

My father and his siblings about the time Patrick Chapman visited the family in the 1940s

Can DNA Tests Detect North American Native Ancestry? What the Science Really Shows

One of the most common questions in genealogy today is whether DNA companies can accurately detect North American Native American ancestry. Many people test hoping to confirm a family story, only to find 0% Native American DNA in their results. Others discover small percentages that raise new questions. And occasionally, a match appears that helps clarify a long‑standing mystery.

That’s exactly what happened in my own research.

Recently, I discovered that I am a DNA match to the grandson of Patrick Chapman, the boy my father remembered visiting the Kapple family in Chicago in the 1940s. According to 23andMe, Patrick’s grandson and I are second cousins once removed. My father had always believed Patrick was a full‑blooded Native American, so I was surprised to see that his grandson had only 2.4% Native American DNA.

That led me to a deeper question:

How good are DNA companies at detecting North American Native ancestry?

The answer is: They can detect it — but only at a broad continental level, and not with the same precision they have for Central and South America.

Here’s why.

Why Central & South American Native ancestry is easier to detect

DNA companies like 23andMe and AncestryDNA have large, well‑sampled reference panels for:

Mexico


Central America


South America

These regions have:

large Indigenous populations


distinct genetic signatures


long‑established communities


many people who have tested

Because of this, companies can identify Central and South American ancestry with high accuracy and often pinpoint specific regions.

Why North American Native ancestry is harder

North American Native ancestry is absolutely detectable — but with limitations. The challenges come from several historical and scientific factors:

1. Most U.S. and Canadian tribes do not participate in DNA testing

Many tribes have:

sovereignty rules

cultural restrictions

privacy concerns

policies against DNA use for identity or enrollment

This means the reference panels for tribes like the Ojibwe, Cree, Menominee, Potawatomi, Lakota, Navajo, and Cherokee are small or nonexistent.

Without large reference samples, companies cannot build precise “tribe‑level” categories.

2. North American tribes historically shared ancestry

Unlike Central and South America — where geography created strong genetic separation — North American tribes:

migrated

intermarried

traded

shared language families

moved with climate and resource changes

This creates overlapping genetic signatures, making it harder to assign ancestry to a specific region or nation.

3. French‑Canadian + Native American ancestry is especially blended

Great Lakes families — including the Chapmans, Masons, LaGrues, and Dufaults — often descend from:

Ojibwe

Cree

Menominee

Huron/Wendat

French voyageurs

Métis communities

These groups intermarried for generations, creating mixed DNA segments that companies can detect as “Native American,” but not as a specific tribe.

So how good are the companies?

23andMe

Very good at detecting small amounts of Native American DNA

Labels it broadly as Indigenous Americas – North

Cannot identify tribe or region

AncestryDNA

Detects Native American ancestry but is more conservative

Often labels it as Indigenous Americas – North, East, or Canada

May miss very small percentages

FamilyTreeDNA / MyHeritage

Detect Native ancestry but with less sensitivity

Sometimes merge North and South American signals
Overall

They can detect North American Native ancestry in most cases — but only at a broad continental level.

How this applies to Patrick Chapman’s grandson

Patrick’s grandson shows 2.4% Native American DNA, which comes from two Ojibwe women who lived seven generations back:

Margaret “Kinikinokwe” LaGrue

Josephte Louise Nischovoise Dufault

Each woman would contribute about 0.78% of DNA at that distance. Because DNA inheritance is random, some people inherit none, while others inherit more than expected.

His 2.4% is exactly what you’d expect from:

two fully Indigenous ancestors

seven generations back

with surviving DNA segments

It also explains why my father’s childhood impression didn’t match the genetics. Patrick wasn’t full Native American — but he did come from a family with deep French‑Canadian and Ojibwe roots, and those roots left a small but detectable trace in his grandson’s DNA.

Why some people with Native ancestors show 0%

Even when the ancestry is real and documented, DNA may not survive through every line. By seven or eight generations back, there is a significant chance that a person inherits no detectable DNA from a specific ancestor.

This is why:

Some people with Native ancestors in the 1700s show 0% Native DNA

Others show 1–3%

A few show higher amounts if multiple Native ancestors are present

Patrick’s grandson is one of the lucky ones — the DNA survived.

Final takeaway

DNA companies can detect North American Native ancestry in most cases, especially when the ancestor is within 5–7 generations. They simply cannot identify the specific tribe or region.

Tracing the Two Mullen Family Houses of Pollaturick: What the 1926 Census Didn’t Tell Me

 

The Cement Era Publishing Company. (1913). The Cement Era: Devoted to cement, concrete and related machinery (Vol. 11). Cement Era Publishing Company.

I went into the 1926 Census with high hopes. After years of trying to untangle the story of the two houses on the old Mullen property in Pollaturick, Milltown, Co. Galway, I thought this census — finally released after a century — might give me the missing pieces. Instead, it handed me a new mystery.

For context, four structures can still be found on the land today:

1. The ruins of what appears to be a piggery

2. Two houses

3. A small structure I originally assumed was a cow house or cart house

My central question has always been simple: How old are the two houses? The answer, however, has been anything but.

What the 1926 Census Actually Says

Unlike the 1901 and 1911 censuses, the 1926 returns give almost no detail about the houses themselves. No construction materials. No notes on whether they were stone, mud, or cement. Just one line that matters:

Two 3‑room houses stood on the Mullen land in 1926.

That’s it.

The acreage, however, was more helpful. It confirmed that Patrick Mullen Sr.’s 57 acres had been divided between his sons Patrick Jr. and Michael, each receiving roughly 28 acres. Both men are listed with 28 acres in 1926, matching the family story and the valuation records.

But the census didn’t tell me which two buildings were the houses — and which were something else.

The Cement House That Never Appears on Any Map

One structure on the property — the mostly cement house near the road — has always bothered me. It never appears on any Ordnance Survey map, even though OS maps are usually meticulous.

Since it’s absent from the 1913–1914 revision, I assumed it must have been built sometime after that survey. Cement cottages were becoming popular in Ireland around that time, promoted as healthier, sturdier replacements for the old stone-and-sod dwellings. The government even offered 60‑year repayment terms to encourage families to upgrade.

So why would a solid cement house be missing from the maps?

To test my assumptions, I asked several AI models. Gemini suggested the house might have been considered a “temporary structure,” and therefore omitted — but that explanation doesn’t hold up. Cement cottages were not temporary. They were seen as modern, sanitary, and permanent. Other AI models agreed: a cement house should have been mapped.

Which brings me back to the original problem:

If the present road‑front house wasn’t there in 1913–1914, where was the second house listed in the 1901 and 1911 censuses?

What the Maps Actually Show

This is a map showing the current location of the houses on the property:

Here’s the timeline as the maps reveal it:

1838 Ordnance Survey (black & white and color versions)

Both show a house on the Mullen land in the same location as the older surviving cottage today

1860s reprint of the 1838 survey

Identical — still one house.

1892 resurvey

More detail, but still no sign of the cement house.

1890s Land Office 25‑inch edition

This map was created for the Land Purchase Acts and is extremely detailed. It shows the main house and a smaller structure behind it. This could be the second house listed on the census records, but it could also be another outbuilding? The the long narrow structure was likely a farm building.

1913–1914 6‑inch revision

This is where things get interesting. The map shows:

• The main house

• A smaller outbuilding or cottage behind it

• A long narrow farm building

• The last 6 inch map (Geohive) now shows the possible cottage beside the house

Below is the last 6 inch version from the Geohive site. The cement house by the road isn't on the map.. Even more puzzling: the smaller cottage beside the main house appears here, but it never appeared on earlier maps. Yet it clearly existed — and still exists.

So the maps themselves are inconsistent

The 1913-1914 6 inch version at the National Library of Scotland doesn't show the cottage.

Could the “Cow House” Actually Have Been the Second Dwelling?

For years I assumed the small structure near the house was a cow house or cart house. It looked utilitarian. It sat at an angle. It wasn’t large.

But then I noticed something that changed everything: It has a large fireplace.

Outbuildings in rural Ireland did not have large fireplaces. Cottages did.

When I asked AI models about this, they unanimously agreed: a fireplace of that size indicates human habitation, not livestock housing.

A neighbor once mentioned that “another house” had stood behind the main house. That comment never made sense to me — until now. There were structures recorded on the maps behind the older house.

If the road‑front cement house wasn’t built yet in 1901 or 1911, then one of the unidentified structures on the maps must have been the second dwelling listed in those censuses.

And the only structures that fit the evidence — maps, ruins, chimney, and census data — is the one I had mistakenly labeled a cow house, and also possibly the small structure behind the house..

So Where Does This Leave the Timeline?

Here’s the most likely sequence based on everything we know:

• Pre‑1838: One house on the property

• 1838–1890s: Still one house

• Between 1890 and 1901: A second small cottage is added beside the main house or behind the house

• 1901 & 1911 censuses: Two households recorded — matching the two cottages

• 1913–1914 OS revision: Three possible cottages appear, and/or farm buildings

• Post‑1914: The cement house near the road is built, replacing the older second cottage

• 1926 census: Two 3‑room houses — likely the main house and the newly built cement house

• Today: The main house survives; the older possible second cottage survives as a ruin; the cement house stands by the road

The “cow house” was almost certainly not a cow house at all. It was a cottage according the AI models.

The Mystery Isn’t Fully Solved — But It’s Finally Taking Shape

The 1926 Census didn’t give me the answers I hoped for, but it forced me to look harder at the physical evidence: the maps, the ruins, the chimneys, the footprints of buildings long gone.

And now, for the first time, the pieces are starting to align.

The little structure I dismissed as a cow house may actually be the key to understanding the entire property — the second home where one branch of the Mullen family lived before the cement house was ever built.

Who Filled Out Irish Census Forms? The System which began in 1841

I didn't know until the other day that the Irish census enumerations were different than the way censuses were enumerated in the United States at the same time. I knew the head of household signed it but didn't know they were also supposed to fill it out. 

Differences between the two countries

In the United States, up to 1960, enumerators filled out the census returns. In Ireland, however, the heads of households were supposed to complete the Census Form A themselves. The other forms were filled out by the constable during British rule, and later by the Garda after independence.

Here are the instructions for collecting the Census A forms for 1926:

I also wondered what would happen if a family didn’t have a pen or ink. Apparently, these could be expensive at the time. The constable or Garda would provide writing materials if needed.

Evaluation of my Mullen Family's forms:

Form B was filled out by the constable or Garda. This form can be useful for comparing handwriting on the census documents to determine whether the head of household actually filled out Form A, or whether someone else did.

In 1901, it appears that the constable filled out the form for my great-great-grandfather, Patrick Mullen, who as head of household was supposed to complete it himself. He could sign his name, but may not have been comfortable enough to fill in the rest of the form.

 

Patrick signs the return but the handwriting in the body of the form is different

This is the Mullen form A in 1911:

The signature on this form is quite different from the one in 1901. Did Patrick sign this? I have doubts. One of the Mullen children may have signed their father’s name. Again, the professional style of the handwriting suggests that the constable filled out the body of the form.

A sample of the constable's writing in 1911:

By 1926, it appears that a family member filled out the census. It could have been his son Michael, or possibly his wife Ellen. Patrick likely signed it:

“Catholic” is misspelled as “Cathlic,” which suggests a household member filled it out. The Garda also appears to have made some corrections. The rest of the handwriting differs from that of the Garda.

Pat Mullin’s signature is clearly different from the handwriting in the body of the form:

Here is a sample of the Garda's writing:

By 1926, there was generally someone in the household able to complete the form. Misspellings like this point to a household member rather than an official completing Form A.

It has been interesting comparing the handwriting on the census forms from 1901, 1911, and 1926. The improvement in the education system is very apparent when looking at these records. Patrick Mullen was born around 1840, while his son was born in the 1880s—a generation that benefited from improved educational opportunities.

It’s a reminder that census records don’t just document who our ancestors were. They also capture how they lived, what they could do, and how the world around them was changing.

The Looooong Wait for the 1926 Census for Ireland Is Over

 

The Long Wait for the 1926 Census

Most countries have a 10‑year wait between census releases. Ireland had a 15‑year gap between the 1911 and 1926 censuses — a long wait. On top of that, census records are restricted for 100 years. In the United States, census records are restricted for 72 years, and the most recent one available to search is the 1950 Census.

Unfortunately, I had an emergency eye exam on the day of the census release. It turned out not to be an emergency — just a vitreous detachment, which can happen with age. I had to wait for the dilating drops to wear off before I could look at the records. Finally, I got to see what I had waited years for.

Searching for Patrick and Mary Mullen

I was searching for my great‑great‑grandparents, Patrick Mullen and Mary Huvane/Huane, in Pollaturick, Milltown, Tuam, Co. Galway, Ireland. I searched for “Mullen” in Pollaturick but didn’t see them on the list that came up. Maybe my eyes were still bad? I eventually found them by searching the townland instead.

What I Found on Form A

I first looked at Form A, which lists the individual members of the household. The first thing that caught my eye was that Patrick and Mary’s son Michael was living with them, along with his wife and three children. He was the eldest son and inherited the property when his father died. His siblings — Helen, Thomas, Bridget (Sister Mary Kathleen), and Winifred — all migrated to the United States.

I knew about Michael Mullen and his wife Ellen Charles’s son Patrick and daughter Mary Ellen, but not their son Aeneas. I’ll have to research Aeneas and his descendants. I’ve been hoping to find family photos that their descendants might have.

I wondered what the notation for Michael and Ellen’s children meant in the “Marriage or Orphaned” column. The notation for the children is “BP IH alive.” I knew it meant both parents were alive, but I didn’t know what “IH” meant. I asked Copilot AI, and it suggested that IH likely stands for “in house.” That does make sense.

The age column caught my attention as well. Ages were supposed to be recorded in years and months. Only two household members have months listed. So were all the others born between March 18 and April 19? Mary Ellen has no months listed, though she should — she was born in June 1921. Her brother Aeneas's age is given as 3 years and 9 months. He was born in July so 9 months would have been correct. 

My great‑great‑grandparents’ ages are way off. Based on Patrick’s age at marriage, he should be about 76 years old. Age discrepancies are common in Irish census records.

The fact that Michael and his wife lived with his parents is supported by other records. My great‑great‑grandmother Mary was present at the birth of Patrick in 1919, her grandson. Mary Ellen was present at the death of her grandmother Mary in 1940.

Mary Ellen Mullen‑Thornton died recently at age 102. She was listed on the 1926 Census.

Trying to Date the Mullen Houses

I was happy to see acreage included on Form A. I’ve been trying to figure out when the front house on the Mullens’ property was built. I was disappointed that the Building Form B didn’t include descriptions of each house like the 1901 and 1911 censuses did. Those descriptions could have helped me determine whether the front house on the Mullen property existed in 1926. Using the ordinance maps, I determined it was likely built after 1913.

Instead, I had to look for clues in the order of visitation on Form B1. Michael Mullen — my ancestor Patrick’s brother — lived in a house set back from the road. The newer house is on the road. My ancestor Patrick is number 146 in the order of visitation; his brother Michael is 152. They both lived on land owned by their father, which they split evenly. Each reported having 28 acres of land. Patrick owned the land near the road. I believe he was visited earlier because his house was easier to reach. But I can’t be certain — the census taker may have had to return to some houses if no one was home on the first visit.

What Form B Does (and Doesn’t) Tell Us

The 1926 Census House and Building Form B doesn’t provide the detailed information about building materials or outbuildings that the 1901 and 1911 censuses included. It only records whether the house was inhabited and how many rooms it had. Both brothers, Michael and Patrick, had the traditional three‑room cottage.

Worth the Wait

I enjoyed seeing my family on the 1926 Census. It was worth the wait.

Overview of My Online RootsTech 2026

I attended RootsTech 2026 online this year. While I enjoyed the experience overall, I was a little disappointed by the lack of major announcements. One of the biggest genealogical events of the year—the release of the 1926 Census of Ireland coming this April—was mentioned only briefly. Still, the sessions on AI were genuinely interesting, especially since I use AI tools regularly in my research.

Exploring My Brick‑Wall Campbell Line with AI

One of the highlights of the conference for me was experimenting with the Copilot AI chatbot on my long‑standing brick wall: my Sarah “Sally” Campbell line. I’ve suspected for a while that she was an aunt or cousin of George Lafayette Campbell. The only evidence so far is DNA—my autosomal matches consistently point toward a Campbell connection, and since I haven’t found a link through any other line, Campbell remains the strongest possibility.

Copilot suggested that Sarah might be the daughter of John Campbell and Elizabeth Dobkins. After digging into that possibility, it didn’t seem to fit. But the AI did surface several promising leads that I hadn’t considered.

One of the most intriguing was the observation that George Lafayette Campbell’s ancestors originally settled in Bedford County, Virginia before migrating to Tennessee. When I checked the 1850 census, I found George living with an older Campbell male born in Virginia. My 3rd great-grandmother Sarah Campbell married Anderson Wray, who was born in Franklin County, Virginia—formerly part of Bedford County. The Wray family in that region also had Campbell connections. All of these overlapping details make the Virginia‑to‑Tennessee-to-Indiana migration path feel increasingly plausible for both families.

Trying Out the New Ancestor Connector Feature

I also explored a new RootsTech feature called Ancestor Connector. It uses your FamilySearch tree to identify connections to individuals mentioned in the Wilford Woodruff Papers. At the top of the page, it claimed I was an 11th cousin three times removed of Wilford Woodruff, though I’m not entirely sure what that’s based on.

The tool scans every name in Woodruff’s writings. As an early Latter‑day Saint leader, he documented many people he encountered—mostly church members, but not exclusively. It appears that some distant cousins of mine may have been mentioned in his papers, though no direct ancestors. Since the feature relies on the shared FamilySearch tree, which contains its fair share of errors, these distant relatives are unlikely to be useful to me at this time.

Since my FamilySearch tree is so bad the Ancestor Connector pointed to Pocahontas and James IV of Scotland as being my ancestors. There is no proof for that at all. 

 Some of my favorite RootsTech 2026 Online Sessions

1. From Chaos to Clarity: Deciding What to Research When Everything Feels Important D. Joshua Taylor
2. DNA Misconceptions Kelli Bergheimer
3. AI & Family History: Foundations & First Steps: Mastering AI Basics for Genealogy Steve Little
4. DNA Evidence Analysis with AI Nicole Dyer
5. DNA Swim School – Part 1: Floating with One DNA Match Diahan Southard
6. DNA Swim School – Part 2: Treading Water with Your Matches Diahan Southard
7. DNA Swim School – Part 3: Swimming Toward Your Ancestors Diahan Southard
8. How Full-Text Search Can Uncover Latin American Genealogy Edward A. Rueda
9. New Ancestor Connector: Part 1 - The Tool Kristy Wheelwright Taylor

As the conference wrapped up, I found myself thinking about how genealogy keeps evolving. Even in a year without big announcements, there are always new tools, new angles, and new ways of seeing old problems. AI isn’t replacing the careful work we do as researchers, but it is becoming a surprisingly helpful companion—one that can surface patterns, point out overlooked connections, and nudge me toward places I might not have thought to look.

My Campbell line has been a brick wall for so long that any hint of movement feels exciting. Whether these Virginia leads pan out or not, they’ve given me a clearer direction and renewed energy to keep digging. And maybe that’s the real value of RootsTech this year: not the splashy announcements, but the reminder that progress often comes from small insights, quiet tools, and the willingness to look again with fresh eyes.

I’m curious to see where these new clues lead next, especially as more records—like the 1926 Ireland census—become available.

I know where I will be on April 18, 2026! Stay tuned!

Can ChatGPT Analyze Your Raw Data Zip file? Yes it can!

 When I was analyzing my new AncestryDNA ethnicity results I wondered if no calls could have affected my ethnicity results? Could the missing ethnicity be due to no calls? I asked ChatGPT that question. I was surprised when I got the answer that I could upload my raw data and check for no calls. I found out I only have 1.64% no calls. According to Chatgpt this is a low percentage of no calls and they are spread throughout my genome. This shouldn't affect my ethnicity results or matches. 

No calls look like this in your DNA raw data:

Correction:

I had to make some corrections to this post. I had actually used a 23andMe raw data file by mistake when I first had ChatGPT search for no calls. I now had ChatGPT search an AncestryDNA file for no calls. I have more no calls at AncestryDNA. According to ChatGPT it has 1.81% no calls. More than 23andMe. I had to tell ChatGPT to look for 00 separated, otherwise it can't find the no calls. Hopefully it did process the file correctly. Eventually I'll figure the percentage myself to verify. Below is what ChatGPT told me about these these no calls and their impact:

🧬 1. DNA Matches (Cousin Matching)

• Small impact overall: The algorithms (at Ancestry, 23andMe, MyHeritage, etc.) look for long stretches of identical DNA (identical-by-descent segments). A missing SNP here and there usually doesn’t break those segments because they use overlapping windows of markers.

• But: If no calls cluster in certain regions, they can slightly shorten detected segments or even cause very small segments to go undetected — potentially reducing the total shared DNA a little.

• In practice, you might lose a few distant matches (like 6th–8th cousins), but your close and moderate matches remain unaffected.

---

🌍 2. Ethnicity Estimates

• Some effect — but minor overall:
  Missing SNPs mean less data for ancestry modeling. However, ethnicity algorithms are designed to tolerate some missingness (they use thousands of reference markers per region).

• Still, a 1.8% missing rate might cause:

  • Slightly less confidence in small trace percentages.

  • Minor shifts in boundaries between similar populations (like between France and Germany, or between Indigenous Americas regions).

• It won’t, however, cause a dramatic change (e.g., losing or gaining a whole continent).

---

🧩 3. Raw Data Uploads to Other Sites

If you upload your Ancestry file to GEDmatch, MyHeritage, or FamilyTreeDNA:

• Some platforms will impute missing data (estimate it from neighboring SNPs).

• Others will just treat those as blank, which can reduce overlap with other users’ files and slightly lower your match count.

---

In short:
✅ No major effect on close matches or main ethnicity regions.
⚠️ Possible small shifts in minor ethnicity regions or very distant matches.

Zero Zero, instead of a letters, means they could not determine your genotype. Apparently AncestryDNA treats no calls as wildcards, and assign a genotype. 

More Analysis from ChatGPT

I decided to ask ChatGPT to analyze my DNA for more ancestry and health related markers. I've checked the markers at 23andMe to make sure they were reported by ChatGPT correctly. ChatGPT seems highly accurate, but you should confirm if a marker is of particular importance to you. I've also uploaded to Gemini AI, but found they didn't always report the genotype correctly (meaning AG, TT, etc.). 

 It can only give very limited information regarding ethnicity. ChatGPT relies on only a few markers for ethnicity related characteristics, such as skin color and hair texture etc.. It doesn't have a reference panel. It would need a reference panel to provide more ancestral information. 

Health information was more detailed. It searches the web for information about your health related markers. The information it provided about my Thyroid disease could prove very helpful. According to one genetic marker I might have a hard time processing the thyroid replacement hormone and may need to change my treatment. Since I have continuing symptoms this might help? 

Ancestral information from ChatGPT:

What the markers show (straight facts from your file)

These are the genotypes I found in your file for the 11 AIMs we checked:

• rs1426654 (SLC24A5) — AA — strongly European (light-skin associated)

• rs16891982 (SLC45A2) — GG — European-associated allele

• rs12913832 (HERC2/OCA2) — AA — European (eye-color) allele

• rs3827760 (EDAR) — AG — heterozygous for the East-Asian/Native-American-associated variant

• rs260690 — AC

• rs4891825 — AA

• rs4798812 — AG

• rs2065160 — AA

• rs4988235 (LCT) — AG — lactase persistence allele present (typical in Europeans)

• rs12203592 (IRF4) — CC

• rs2814778 (DARC / Duffy) — TT — not the African Duffy-null allele

Short interpretation of those genotypes:

• Strong overall European signal — multiple canonical European alleles are homozygous or present (SLC24A5, SLC45A2, HERC2, LCT). That fits a French / Northwestern European ancestry profile.

• One heterozygous EDAR (rs3827760 AG) — this is the interesting bit. The derived EDAR allele is common in East Asian and Native American populations and rare in most Europeans. A single copy (AG) suggests a small contribution somewhere in your ancestry that carries that variant — consistent with limited Native American or East-Asian-related ancestry, which is plausible in some French-Canadian pedigrees (due to historical admixture) or could come from other branches.
  — Important: one heterozygous EDAR does not imply large East-Asian ancestry; it’s just an informative signal that merits follow-up.

• No signals of significant recent African ancestry in the small marker set (rs2814778 not showing the African-null allele).

I definitely have Native American ancestry, so that information is correct. 

What ChatGPT Can Tell You About Your Raw DNA Data (information from ChatGPT on what it can do)

Beyond ancestry reports and trait predictions offered by these companies, AI tools like ChatGPT can help you interpret your data in fascinating ways.

1. Understanding Your Genetic Variants
Raw DNA data is essentially a long list of genetic variants, called SNPs (single nucleotide polymorphisms). Each SNP represents a small change in your DNA that can affect traits, ancestry, or disease risk. ChatGPT can help you understand what specific variants mean by explaining:

• How a particular SNP has been associated with health traits in scientific studies

• How common the variant is in different populations

• Whether it might contribute to certain traits like eye color, lactose tolerance, or even caffeine metabolism

2. Health Insights (With Caution!)
While ChatGPT can summarize research about genetic variants, it cannot provide medical advice or diagnoses. However, it can:

• Identify genetic markers associated with certain conditions, such as thyroid disorders, cardiovascular traits, or metabolic differences

• Explain the current state of research on a gene or variant

• Suggest areas you might discuss with a healthcare professional

It’s crucial to remember that genetics is only part of the story. Lifestyle, environment, and other factors often play a larger role than a single genetic variant.

3. Ancestry and Ethnicity Clues
Even though raw DNA files don’t come with polished ancestry reports, ChatGPT can help interpret markers linked to specific populations or regions. For example:

• Identify SNPs more common in Native American, European, or African populations

• Suggest possible regional connections based on genetic patterns

• Help you explore historical migrations reflected in your DNA

4. Traits and Personal Characteristics
ChatGPT can summarize associations between your DNA and non-health traits, such as:

• Hair color, eye color, and other physical features

• Taste preferences or sensitivities (like bitter taste perception or lactose intolerance)

• Sleep patterns or exercise response

5. Research and Curiosity
Your raw DNA is a powerful tool for personal exploration. ChatGPT can guide you in:

• Researching genes or variants you find interesting

• Comparing your genetic profile to published studies

• Exploring ancestry in a broader historical and geographical context

6. Limitations
It’s important to approach AI interpretations carefully:

• Not all genetic associations are strong or clinically relevant

• ChatGPT relies on publicly available research and cannot access private medical records

• Always consult a healthcare professional for medical concerns

Conclusion

ChatGPT can turn your raw DNA data into a source of insight and curiosity. From ancestry clues to understanding your genetic markers, AI provides a way to explore your DNA beyond what consumer testing companies offer. While it doesn’t replace professional medical advice, it opens doors to learning more about yourself, your heritage, and the science of genetics.

AncestryDNA Ethnicity Update 2025: Lessons from My Cousins’ Results

 Why My AncestryDNA Ethnicity Results Are Less Helpful Than My Family’s

My AncestryDNA ethnicity results have always been the least helpful of anyone in my family. I believe it’s because I’m more ethnically mixed than most of my relatives. I have some less common ancestry from Burgenland and Nicaragua, along with more typical Western European roots.

I do think it’s interesting to see how accurate AncestryDNA’s regional breakdowns have become. I know my ancestry going back at least a couple hundred years, and so far, I haven’t found any non-paternity events. I’m sure if I traced the lines further back, I’d eventually run into a few, which is why I expect the ethnicity estimates to reflect more recent ancestry. If they reflect ancestry from farther back, it’s harder for me to confirm whether it's accurate or not since I don’t have the documentation to support it. This is especially true when it comes to Irish genealogy — I can only trace those lines back a couple hundred years with records.

For someone who’s adopted, these ethnicity estimates can provide real clues about their heritage. They can also be useful for the descendants of enslaved people. If the estimates continue to improve in accuracy over time, they may become even more valuable for people in those situations. Personally, I already have a good understanding of my ethnic background, so these results aren’t critical for me.

In my experience, looking at the family trees of close DNA matches actually gives a clearer picture of your ethnic origins. These matches can sometimes pinpoint your ancestors’ exact towns or villages. You can look for recurring place names among your matches to help narrow things down. AncestryDNA Journeys is another tool that can point you in the right direction. Journeys are actually more useful for recent ancestry.

Paternal First Cousins Comparisons

So, why are my first cousin’s DNA ethnicity results more helpful than mine?

His results pinpoint the exact area where our paternal great-grandmother, Helen Mullen Mason’s, recent ancestors came from. They were all from Connacht, Ireland — specifically around Pollaturick, Milltown, Tuam, Co. Galway — for several generations before her birth. Unfortunately, I haven’t been able to trace this branch of the family past 1800. Our Irish surnames include Mullen, McQualter, Huvane/Huane, and Rattigan. I do know that the Huvane/Huane name goes back quite far in the Kilvine area of Co. Mayo.

My paternal first cousin has 9% Connacht Irish ethnicity, which perfectly matches what we know about our great-grandmother Helen Mullen Mason’s background.

I have 2% ancestry from Connacht, along with a mix from other regions, and 9% from Leinster. That doesn’t seem to make sense. Perhaps I had ancestors from there further back in time, but since my cousins’ results are so different, I doubt it. We don’t share the exact same DNA segments, and even so, the Leinster percentage seems questionable. 

It appears some of my Leinster is from my mother's side, but the highest percentage is from my father's side. Ireland being a small island it's likely that I have ancestry that traces back to many areas in Ireland. The Leinster result isn't very useful if someone is looking for hints regarding more recent Irish ancestry. 

Here are the percentages for all 3 of my paternal first cousins who tested at AncestryDNA. All have a higher percentage of Connacht DNA. I have 2%. If I didn't know my ancestral origins already I might guess my Irish ancestors were recently from Leinster, Ireland. 

Why Does AncestryDNA Show Zero French or French Canadian DNA—Even Though We Have a French Canadian Great-Great-Grandfather?

One of the biggest surprises in our DNA results is that none of us show any French or French Canadian ethnicity on AncestryDNA, even though we know — without a doubt — that our great-great-grandfather was French Canadian. This isn’t a case of a misidentified ancestor or a non-paternity event. We've confirmed the connection through solid DNA matches on that line.

The first result in the row is my result, the second is my mother's. She has no French recent ancestry. The next three rows belong to my paternal first cousins. 

French Canada has been added to regions, specifically Quebec. None of my family members have this admixture. 

So, why is AncestryDNA missing this?

1. Ethnicity estimates are not exact

AncestryDNA ethnicity results are just that — estimates. They're based on comparing your DNA to reference panels of modern populations, and that means there’s always some margin of error or generalization. Labels like “French” or “Quebec” may not reflect exactly how your DNA is inherited across generations.

2. DNA inheritance is random

You inherit about 3.125% of your DNA from a single great-great-grandparent. Because of random recombination, it’s entirely possible that the small DNA segments you received from that French Canadian ancestor don’t contain markers that Ancestry identifies as French or Quebec-related. That doesn't mean the ancestry isn't real — it just didn't get passed down in a way that Ancestry’s algorithms detect.

3. Quebec is now a defined region — but you may not meet the threshold

AncestryDNA now includes a defined region for Quebec, which is a great improvement. However, not everyone with French Canadian ancestry will be assigned this region, especially if that ancestry comes from just one great-great-grandparent. The region tends to show up more clearly for people with multiple lines going back to Quebec, or more recent Quebecois ancestry. A single ancestor four generations back may simply not leave enough detectable DNA for Ancestry to flag the region.

4. French and French Canadian DNA overlaps with nearby populations

Genetically, French ancestry — especially from the early settlers of Quebec — often overlaps with populations in Belgium, Northern France, Western Germany, and even parts of the British Isles. French Canadian settlers came from various regions of France, many of which share DNA markers with other Western European populations, making it harder to isolate them as distinctly “French” in ethnicity reports.

5. Reference populations still have limitations

Even with the new Quebec region, Ancestry's ethnicity algorithm is only as good as its reference populations and sampling. While the database is growing, it may still lack the resolution to detect subtle French Canadian DNA in someone with limited recent ancestry from that region — especially if it's diluted over several generations.

---

So, while our great-great-grandfather was clearly French Canadian — and we have both the family records and DNA matches to support it — AncestryDNA’s ethnicity estimate doesn’t reflect that, and that’s not unusual. Ethnicity results can be interesting, but they’re only a rough sketch. For deeper insights, looking at shared matches, family trees, and regional migration histories usually tells a more accurate story.

The ethnicity results are much better than they were when these tests were first offered. They still aren't accurate enough to paint a complete picture of your ethnic roots. 

Finding Winifred Mullen: A Long-Lost Aunt, a Boy from Mackinac Island, and an Unexpected Family Connection

My father Robert Kapple standing behind his sister around 1940 (about when Patrick Chapman visited). On right, you can see how much information I've collected on Winifred Mullen since earlier this month.

Winifred Mullen Found!

For years, my Great-Great Aunt Winifred Mullen was just a name in old Irish census records—born around 1892, living in Ireland in 1901 and 1911. I assumed she had stayed there her whole life, unlike her siblings—my great-grandmother Helen Mullen-Mason, her sister Bridget (who became Sister Mary Kathleen), and their brother Thomas—who all immigrated to Chicago (their brother Michael Mullen remained in Ireland until his death).

But Winifred had a secret: she, too, had crossed the Atlantic.

The Ellis Island Breakthrough

After years of searching Irish records for a marriage or death certificate, I finally found her—on a passenger list. In 1913, she arrived at Ellis Island, her name misspelled so badly that previous searches had missed her. Like her siblings, she was headed to Chicago.

But then… she vanished again.

The Chicago Clue

Frustrated, I turned to Ancestry family trees—usually a last resort due to unsourced information—and there she was: Winifred Mullen, married to Henry Chapman in Chicago in 1929. No parents listed, but the details fit (but her married name, Chapman, led me to find more records):

• By 1930, they were living on Mackinac Island, Michigan.

• The 1930 Census gave her year of immigration as 1913 (oddly her marriage record didn't come up in the searches until I included the name Chapman)

• They had a son, Patrick Henry Chapman, born in 1931 

And then I remembered something my father had mentioned decades ago…

The Little Boy from Mackinac Island

My father, born in 1933, once recalled a childhood visit from a young relative—a quiet boy from Mackinac Island, around his age. The boy, he said, was Native American, which made him wonder if we had Indigenous ancestry.

The family thought the boy hadn’t enjoyed his visit—until they heard he couldn’t stop talking about it when he got home.

Now, the pieces fell into place: That boy was Patrick Chapman, Winifred’s son.

The Métis Connection

Further research revealed that Henry Chapman’s family had Métis roots—descendants of French-Canadian fur traders and Ojibwe women on Mackinac Island. Patrick, as a tribal member, carried that heritage.

But our link to him wasn’t through his father—it was through his Irish mother, my long-lost aunt.

A Bittersweet Discovery

Winifred died of breast cancer in her 50s. Patrick passed in 2002, and his only daughter is also gone. I’ll never hear his memories of visiting Chicago or what he thought of my father’s family.

But this journey taught me something: Family stories hold hidden truths. A passing comment about a "Native American cousin" led me to an entire branch of my family tree—one that tied Ireland, Chicago, and Mackinac Island together.

If you’ve ever hit a brick wall in genealogy, don’t give up. Sometimes the answer lies in:

• Misspelled names (check every variation!)

• Unexpected locations (who knew Winifred went to Mackinac Island?)

• Family stories (even vague ones can hold clues)

• Indexing and Transcription Errors (errors in indexing can affect search results)

And if you have ties to the Chapman family of Mackinac Island, I’d love to hear from you. There are still more stories to uncover, and photos.