.png?format=1500w)
Understanding Genetic Genealogy and Direct-to-Consumer DNA Testing
I have spent years working with people trying to make sense of their raw DNA data files and match lists from various commercial testing companies. The general process sounds straightforward but in practice it involves a lot of quiet frustration, some bad assumptions, and a few technical tricks that never get explained properly in the marketing copy. The phrase itself appears in search results as a curiosity-driven headline that blends celebrity interest with genetic genealogy buzzwords. There is no publicly available, credible scientific analysis of any such "DNA mystery" from Britney Griffin, the American former gymnast. What actually exists are routine genetic genealogy tools, population genetics databases, and direct-to-consumer genotyping services that anyone can use if they understand how they work and where they fall apart. I will walk through the real mechanics here instead.
How Commercial DNA Testing Actually Works
Companies like AncestryDNA, 23andMe, MyHeritage, and FamilyTreeDNA use SNP microarray chips to genotype roughly 600,000 to 2,000,000 single nucleotide polymorphisms across the genome. This is not whole genome sequencing. It is a targeted snapshot. The chip asks a finite set of predetermined questions and records your answers at those positions. Everything between those positions is inferred or left completely blank. Your saliva sample arrives at a lab. DNA is extracted, amplified, fragmented, and hybridized to the microarray. The machine reads fluorescent signals and converts them into genotype calls: homozygous reference, heterozygous, or homozygous alternate at each SNP position. Those calls are then compared against the company's reference population panels to estimate ancestry composition, and against the company's internal database of other tested users to identify shared DNA segments and predict relationships. The ancestry estimates come from reference panels that are themselves imperfect. If your heritage falls outside the populations the company has sampled, the algorithm will force your DNA into the closest available category and report a confidence level that sounds definitive but often is not. I have seen cases where people with significant North African or Middle Eastern ancestry received predominantly Southern European estimates because the reference panel simply did not have enough coverage from those regions at the time of testing.
Reading Your Raw Data File
Every major testing company allows you to download a raw data file. It is typically a tab or comma delimited text file containing chromosome number, genetic position, and genotype calls for hundreds of thousands of SNPs. The format varies slightly between providers but follows a broadly similar structure. You can upload this file to third-party platforms for additional analysis. GEDmatch, DNA.Land, and various academic tools accept these files. The value here is that you are not locked into one company's interpretation of your data. Different algorithms, different reference panels, different segment detection methods. Running your raw data through multiple tools often reveals discrepancies that are worth investigating rather than ignoring. One practical thing most beginners miss: raw data files use different build assemblies. AncestryDNA historically used GRCh37/hg19 coordinates while some other platforms may use GRCh38/hg38. Uploading a file in the wrong build to a tool that expects the other will produce garbage results. Always verify which assembly a tool requires before uploading. I once spent six hours debugging what I thought was a complex pedigree issue before realizing my coordinates were off by roughly 100,000 positions because I had mixed up the builds.
Get the Full Details
DNA Matching and Segment Analysis
When two people share a segment of DNA, it means they inherited that segment from a common ancestor. The length of the shared segment in centimorgans (cM) gives you a rough estimate of how recently that ancestor lived. Longer segments indicate closer relationships. A full sibling share will typically be around 2,600 cM across many segments. A third cousin might share 50 to 200 cM spread across a handful of segments. The Shared cM Project database, maintained by genetic genealogists, provides empirical data on how much DNA various relationships actually share. It is one of the most useful open resources in this field. You can look up any predicted relationship and see the observed range. I rely on it constantly when evaluating whether a match makes sense given the paper trail I have built. Segment data also reveals something that is easy to overlook. Identical by descent (IBD) segments are inherited from a real common ancestor. Identical by state (IBS) segments are coincidental matches that happen because certain alleles are common in the population. Most consumer platforms filter out IBS segments automatically but not always perfectly. When you are working with distant matches in the 5 to 20 cM range, IBS contamination becomes a real concern. I have found that cross-referencing matches on multiple platforms and requiring triangulation with a third party who shares the same segment dramatically reduces false positives.
Troubleshooting Bad Matches and Unexpected Results
Not every match list behaves the way you expect. Endogamy is one of the biggest sources of confusion. In populations where ancestors married within a relatively small community for many generations, descendants tend to share DNA across many chromosomes rather than concentrated in a few segments. This makes relationships appear closer than they actually are. If your match list is flooded with matches showing unexpectedly high cM values but no paper trail connecting you, endogamy is a likely explanation. The Avicenna, Ashkenazi Jewish, and Acadian populations are well-documented examples where this effect is strong. Another frequent issue is segment threshold settings. Platforms default to a minimum segment size, usually around 5 to 7 cM, but lowering that threshold increases sensitivity and also increases noise. I typically recommend starting with a 7 cM threshold and 1 cM SNP count minimum for standard analysis, then relaxing those parameters only when you need to investigate a specific hypothesis. Phasing errors can also produce ghost segments. Phasing is the process of determining which alleles on a chromosome came from which parent. If the phasing is incorrect, the algorithm may split a genuine inherited segment into two smaller fragments or merge segments that should be separate. This mostly affects very small segments below 10 cM and is less of a problem for close relatives where the signal is strong.
Limitations That Nobody Wants to Advertise
Commercial DNA tests cannot determine your exact ethnicity. They can estimate it based on reference populations, and those estimates change as the reference panels grow and the algorithms are updated. I have watched ancestry compositions shift noticeably between test updates for the same person. A 2021 result might show 12% British and Irish while the same person tested again in 2024 shows 8% because the reference panel was refined. Neither result is definitively right or wrong. Both are point estimates subject to sampling error. DNA testing also cannot solve all genealogical problems. If a branch of your family tree has missing records, lost surnames, or adoption events with no documented information, DNA can narrow the possibilities but it cannot generate records that do not exist. The combination of DNA and traditional genealogy is necessary. DNA alone is incomplete. Traditional genealogy alone is also incomplete. Together they are far more powerful than either by itself. There is also the privacy dimension. Once you upload your raw data or create an account with a testing company, your genetic information exists in their database. Law enforcement access, data breaches, and policy changes are real considerations. Some people choose to test and keep their data private. Others upload to multiple platforms and accept the broader exposure. There is no universally correct answer here.
Practical Workflow for Serious Research
Start with a test from the platform with the largest user base if your goal is finding matches. AncestryDNA currently has the largest database in North America. Upload your raw data to GEDmatch and any other third-party tools you find useful. Build a proper pedigree alongside your DNA work. Document every match, every shared segment, and every hypothesis you test. Keep notes that you can revisit months later when you need context. When you hit a dead end, which you will, switch strategies rather than grinding the same approach. Try clustering matches using the Leeds Method or the chromosome browsing tool to identify groups of people who likely share a common ancestor. Then work backward from those clusters to find the connecting node in your family tree. The process is rarely linear. It involves patience, careful note-taking, and a willingness to update your conclusions when new data arrives. The best results come from people who treat DNA analysis as a research discipline rather than a quick lookup.
