Gene Testing and Identity Reconstructions
The whole industry around genetic ancestry reports has gotten pretty noisy over the last few years. People upload raw DNA data, cross-reference with public datasets, and start building family narratives that are only as solid as the reference panels they're pulling from. It is a messy process and most people do not realize how much of it is basically educated guessing with extra steps. I ran into this exact situation when someone reached out asking about a Rain Spencer situation. The public record was thin, the family history was fragmented, and the only real data point available was a consumer genetic test result. What happened next is the kind of thing that comes up more often than you would think. Here is how I approached it.
First, I pulled the raw DNA data from whatever upload source was available — usually 23andMe or AncestryDNA. Those files are typically in .txt or .csv format with rsID markers and genotype calls. Once I had that, I ran it through a tool like GEDmatch or Promethease to get ethnicity estimates and match lists. That gives you a starting picture, but the ethnicity estimates are where most people go wrong. They treat percentage breakdowns like hard facts instead of what they actually are: probabilistic snapshots based on whatever reference populations the lab happened to sample at the time. The reference panel issue is the biggest blind spot. A 23andMe ethnicity estimate for "British and Irish" looks very different from AncestryDNA's version of the same category, even when run on the same sample. That alone can shift someone's entire understanding of where their roots supposedly lie. I have seen people panic over a 3% fluctuation that was just a rounding difference between two algorithms. So after the initial upload, I cross-referenced the match lists against public family trees on GEDmatch and MyHeritage. The goal was not to chase ethnicity percentages but to triangulate actual genealogical connections. Shared centimorgan ranges matter more than anything the algorithm spits out for ethnicity. If two people share 85 cM across multiple segments, they are likely second or third cousins regardless of what the ethnicity breakdown says. That practical step — working backward from genetic matches into documented trees — is where the actual work happens.
With the Spencer case, the genetic data pointed toward a mixed heritage profile with some ambiguity around West African and Southern European ancestry. The ethnicity estimates alone were inconclusive because the individual was underrepresented in the reference panels for both regions. What cracked it open was a combination of shared DNA segments with verified third cousins who had paper trails going back to the 1800s, plus a targeted search of archived census and immigration records that lined up with the genetic hypothesis. It took about three weeks of scattered research. The genetic data got us close, but the paper records confirmed what was actually happening. If you are trying to do something similar, the best practical approach is straightforward. Download your raw data from every platform you have an account with, because each company uses a different genotyping chip and you will capture more markers that way. Run the data through a second or third ethnicity estimator to spot inconsistencies — when two different tools give you wildly different regional breakdowns, flag that and dig deeper. Then focus on the shared matches and build out those branches systematically. Document everything as you go because you will forget which tree led where. There are real limitations here that nobody likes to talk about. Small reference populations mean small sample sizes, which means unreliable estimates for people with complex or non-European backgrounds. Endogamous populations like Ashkenazi Jewish or Filipino communities show inflated ethnicity percentages because the reference groups overlap heavily. And then there is the privacy question — once your raw DNA is on a public platform like GEDmatch, it is out there. Law enforcement access, distant relative discoveries, data breaches. All of it is a real factor.
Get the Full Details
The alternative to going down this rabbit hole is hiring a professional genetic genealogist. It costs between $150 and $400 depending on complexity, but someone who does this full-time will save you the weeks of fumbling through mismatched data. I usually recommend that route for anything that feels stuck or produces contradictory results. The DIY method works fine for straightforward cases with strong paper trails already in place. When things get complicated, professional help pays for itself quickly. The bottom line is that ethnicity estimates from consumer tests are a starting point, not a destination. They are useful for narrowing down where to look in public records and match lists, but they should never be treated as definitive. The actual story is in the shared DNA segments, the documented trees, and the paper trail that connects them. Without that verification layer, you are just reading algorithmic noise and calling it heritage.
