Our primer article introduced autosomal DNA (atDNA) testing — what it is, how it works, and which companies offer it. This companion piece goes deeper: into how different companies actually test your DNA and why that matters, how to combine results across platforms to get more from your testing investment, what third-party tools can do that no single company offers, and where atDNA testing has real limitations that experienced researchers need to understand clearly. If you’ve already taken one or more atDNA tests and want to get more from them, this article is for you.
Not All Tests Are the Same: Chips, SNPs, and Genome Coverage
A common misconception is that all autosomal DNA tests measure the same thing. They don’t. Each testing company uses a microarray chip — a laboratory tool that reads your DNA at hundreds of thousands of specific locations across the genome — but those chips differ in which locations they examine, how many they cover, and how that data is processed.
A microarray chip works by probing specific positions on the genome called single nucleotide polymorphisms (SNPs, pronounced “snips”) — locations where the DNA sequence commonly varies between individuals. Most consumer genealogy chips test between 600,000 and 750,000 SNPs[1], which sounds like a large number but represents only approximately 0.02% of your total 6.4 billion base pair genome. What varies significantly is which positions each company tests.
AncestryDNA uses its own proprietary chip design, testing approximately 664,000–700,000 SNPs. Because Ancestry designed its chip independently rather than building on a standard platform, its SNP positions overlap only partially with other companies — roughly 44% overlap with FamilyTreeDNA[2] and MyHeritage, and as little as 14% overlap with 23andMe’s current chip.
23andMe uses its version 5 chip, built on the Illumina Global Screening Array (GSA) platform with custom additions, testing approximately 640,000 SNPs. Because 23andMe and Living DNA both use GSA-based chips, they share over 90% of their SNP positions with each other — but only about 14% with AncestryDNA’s proprietary chip.
FamilyTreeDNA (FTDNA) and MyHeritage both use GSA-based chips with custom additions and share nearly identical SNP sets with each other — and about 44% overlap with AncestryDNA.
MyHeritage introduced Whole Genome Sequencing (WGS) for new kits beginning in late 2025, representing a fundamentally different approach: rather than probing selected positions, WGS reads the entire genome (6.4 billion base pairs) at approximately 30x coverage — meaning each position is read roughly 30 times for accuracy. This is far more comprehensive than chip-based genotyping and eliminates the SNP selection question entirely, though it introduces its own processing and imputation considerations for genealogical matching.
Living DNA uses a customized Affymetrix chip designed specifically to provide mid-range overlap with both the AncestryDNA chip and the GSA-based chips used by other companies — a deliberate design choice intended to improve compatibility when raw DNA files are shared across platforms.
Why Chip Differences Matter for Genealogy
The practical consequence of these differences is that when a raw DNA file from one company is uploaded to another platform, the receiving platform may have to work with a file that covers only a subset of the positions it normally uses. This can affect matching accuracy in two ways:
Fewer shared SNPs means less matching power. If Company A tests positions 1, 2, 3, 4, and 5, and Company B tests positions 3, 4, 5, 6, and 7, a file from Company A uploaded to Company B’s platform can only be matched at positions 3, 4, and 5. Segments that would have been detectable using all of Company B’s positions may be missed entirely, or match thresholds may not be met.
Imputation fills gaps but introduces uncertainty. Some platforms use a statistical technique called imputation to estimate the values at positions not covered by an uploaded file. Imputation uses population data to make educated guesses about what your DNA likely reads at untested positions. This can improve matching performance but introduces a layer of statistical estimation that isn’t present in a natively tested kit. MyHeritage, for example, used imputation when processing uploaded files — one reason why native MyHeritage tests often produced slightly different match results than uploads of the same person’s DNA from AncestryDNA.
This is also why FamilyTreeDNA historically had difficulty accepting 23andMe uploads: the low overlap between Ancestry-style chips and GSA chips meant the uploaded data was too sparse for reliable matching without heavy imputation.
🧬 Building a “Super Test”: Combining Multiple Companies
Because each testing company’s chip covers somewhat different positions across the genome, testing at multiple companies — or uploading your raw data to platforms that combine results — gives you access to a broader, more complete picture of your DNA than any single test provides. Think of it as each company reading a different subset of pages from the same book: combining the results gets you closer to reading the whole thing.
The most powerful way to take advantage of this is through GEDmatch, which accepts raw DNA uploads from all major testing companies and applies consistent analytical tools across all kits, regardless of their origin. When you upload kits from, say, AncestryDNA and 23andMe to GEDmatch, you can compare them against the full GEDmatch database and use segment-level tools — One-to-One comparison, Segment Search, and Triangulation — that draw on whichever kit provides the best coverage for a given region of the genome.
A more advanced approach is to upload raw files from multiple tests of the same person to a tool like DNA Kit Studio (available at dna.land and similar platforms) or use GEDmatch’s own kit management to maintain multiple kits per person from different companies. Researchers who have tested at both AncestryDNA and 23andMe, for example, can upload both raw files to GEDmatch and compare them against the database separately — different matches may appear in each comparison, depending on which chip’s coverage is more relevant for a given segment.
The cumulative effect is meaningful. Researchers who have tested at AncestryDNA, uploaded to FTDNA, and maintained a GEDmatch kit effectively have access to three separate matching databases, each drawing on partially different genomic coverage — substantially increasing the chance of finding a critical match that any single platform alone would miss.
Uploading Raw DNA: The Current Landscape
The ability to download your raw DNA from one company and upload it to another is one of the most cost-effective strategies in genetic genealogy — but the landscape has changed meaningfully in recent years and continues to evolve.
FamilyTreeDNA accepts raw DNA uploads from AncestryDNA, 23andMe, MyHeritage, and Living DNA. A one-time fee of $19 unlocks the full suite of Family Finder matching and analysis tools, including the chromosome browser and advanced matching features. For researchers who have already tested at AncestryDNA, uploading to FTDNA is an immediate way to access FTDNA’s matching database and surname projects — including the Dillman surname project — without purchasing a second test.
GEDmatch accepts uploads from all major companies and remains the premier cross-platform matching tool. Basic tools are free; Tier 1 advanced tools (Segment Search, Triangulation, Multiple Kit Analysis) are $10/month. As detailed in the DFA’s GEDmatch article, this is where the “super test” concept is most fully realized.
Living DNA accepts raw DNA uploads and provides basic matching and ancestry analysis. The platform’s deliberately designed chip overlap with both AncestryDNA and GSA-based chips makes uploaded files particularly compatible.
MyHeritage discontinued accepting raw DNA uploads in August 2025. Access to MyHeritage’s matching database now requires purchasing a MyHeritage kit directly.
AncestryDNA and 23andMe do not accept uploads from other companies and have never done so. Both allow you to download your raw data for use elsewhere.
The Lazarus Tool: Reconstructing a Deceased Ancestor[3]‘s DNA
One of the most remarkable applications of atDNA in genealogy addresses a problem every researcher eventually faces: a key ancestor died before DNA testing existed, and there is no surviving sample. The Lazarus tool at GEDmatch (a Tier 1 feature) offers a partial solution — reconstructing an approximation of a deceased person’s DNA from the tested DNA of their descendants and relatives.
The concept is elegantly simple. Because each child inherits roughly half[4] of each parent’s DNA, a parent’s genome is distributed — in different combinations — across their children. If you test multiple children of a deceased ancestor, the segments they share with each other represent DNA that came from that ancestor (or the ancestor’s spouse). By pooling those shared segments and eliminating contributions from the other parent’s side, Lazarus assembles a “pseudo-kit” that approximates the ancestor’s genome.
The tool works with two groups of relatives:
- Group 1 — Descendants only: Children, grandchildren, or other direct descendants of the target ancestor. Up to 10 kits can be included. Grandchildren are less useful than children since they carry only about 25% of the ancestor’s DNA each, and including a child and their child together may actually reduce quality. Cousins, siblings, and other non-descendants must never go in Group 1.
- Group 2 — Non-descendant relatives: Siblings, parents, cousins, or other relatives of the target ancestor who are not also their descendants. These help confirm that a shared segment actually came from the target ancestor rather than from the ancestor’s spouse. Up to 100 kits can be included.
The resulting pseudo-kit can then be uploaded to GEDmatch and used like any other kit — running One-to-Many comparisons, finding matches, and using it as a surrogate for the deceased ancestor in further analysis. The quality of the reconstruction depends heavily on how many descendants and relatives have tested, and how closely related they are to the target. A reconstruction built from four or more children of the target ancestor is far more complete than one built from grandchildren alone.
As a practical example: if multiple known Dillman descendants of a common great-grandparent have tested and uploaded to GEDmatch, a Lazarus kit can be created for that great-grandparent — giving the Dillman Family Association (DFA) an approximate DNA profile for an ancestor who may have lived in the mid-1800s, which can then be compared against the GEDmatch database to find related lines that documentary research alone would never reach.
Other Advanced Third-Party Tools
DNA Painter ([5][5] target=”_blank” rel=”noopener noreferrer”>dnapainter.com) — Allows you to visually map DNA segments onto a chromosome painting, coloring each segment by the ancestral line it has been assigned to. Over time, as you identify which match’s segments came from which ancestor, you build a color-coded map of your genome. This is one of the most powerful visual tools in genetic genealogy and integrates data from any testing company. The Shared cM Project tool, also at DNA Painter, helps interpret the range of possible relationships for any given shared cM amount.
Clustering tools (AutoClusters, Leeds Method) — Both MyHeritage’s built-in AutoClusters feature and the Leeds Method (a manual technique developed by Dana Leeds) group your DNA matches into clusters of people who all share DNA with each other — typically indicating that they share a common ancestral couple. Clustering is one of the most effective ways to organize a large match list and begin identifying which cluster belongs to which branch of your family tree. MyHeritage provides AutoClusters natively; third-party tools like the Shared Clustering tool (available on GitHub) can apply the same approach to AncestryDNA match data.
Genome Mate Pro — A free desktop application for organizing and cross-referencing DNA match data from multiple companies. Researchers who work across several platforms find it invaluable for keeping match notes, segment data, and correspondence organized in a single local database.
GEDmatch’s “My Evil Twin” Tool — A Tier 1 GEDmatch feature that generates a “mirror” kit showing the DNA you didn’t inherit from a parent (i.e., the half that went to a sibling instead). When a parent has tested, this can help researchers understand which segments came from which grandparent — a useful phasing technique when grandparent kits are not available.
Honest Pitfalls of Autosomal DNA Testing
The database boundary problem. Every testing company’s matching database is a closed system. AncestryDNA can only match you against other AncestryDNA testers; FTDNA matches only FTDNA testers. The cousin who holds the key to your brick wall may have tested at a different company — one you haven’t uploaded to. This is the single most important argument for uploading your raw data to as many platforms as accept it, and for having family members who test at different companies also upload to GEDmatch.
Ethnicity estimates are not genealogical proof. Ethnicity percentages from any company are population-level statistical estimates, not precise measurements of your ancestry. The same raw DNA will produce different estimates at different companies, and those estimates shift when companies update their reference populations. They are a useful starting point for broad-strokes research direction, but they should never be used as evidence for a specific ancestral claim, and they should never be shared without the caveat that they are estimates.
The pileup region problem. Certain regions of the genome have DNA sequences that are nearly identical across large portions of the population — not because of shared recent ancestry but because of ancient population history. Matches in these “pileup regions” often appear to be genealogically meaningful but actually reflect ancient common ancestry shared by nearly everyone of a given ethnic background. GEDmatch’s tools can help identify these regions; experienced researchers learn to be skeptical of matches where the shared segment falls entirely within a known pileup region.
Endogamy. In populations where marriage within the same community was common for many generations — Ashkenazi Jewish communities, island populations, certain religious groups, and others — DNA matches are numerous but difficult to interpret. Because everyone in an endogamous population shares small amounts of DNA with virtually everyone else through multiple ancient pathways, standard relationship estimates and matching thresholds become unreliable. Working with endogamous populations requires specialized techniques beyond the scope of this article.
The disappearing ancestor problem. Because DNA is inherited randomly through recombination, it is possible — and increasingly likely beyond six or seven generations — to share no detectable DNA with a specific ancestor even though that person is genuinely in your genealogical tree. An atDNA test cannot prove that an ancestor is not in your tree; it can only tell you that no detectable shared DNA has survived to the present generation. For testing very deep ancestry, Y-DNA and mtDNA are far more reliable on their respective direct lines.
Relationship ambiguity. The same amount of shared DNA — say, 200 cM — is consistent with multiple different relationships: half-first cousin, first cousin twice removed, half-great-aunt, and several others. DNA alone cannot distinguish between these possibilities; documentary research is always needed to determine the actual relationship from the range of possibilities the DNA suggests. The Shared cM Project tool at DNA Painter makes this ambiguity explicit and is an essential reference for any researcher working with match relationships.
Privacy considerations. Taking a DNA test doesn’t only reveal information about yourself — it reveals information about all of your biological relatives, including those who have not consented to testing. Before testing, consider whether any close relatives might be surprised or distressed by discoveries the test could reveal, such as unexpected parentage relationships, unknown adoptions, or half-siblings. These discoveries happen more often than most people expect, and having thought through how you would handle them in advance is genuinely worthwhile.
A Strategy for Serious Researchers
Putting it all together, here is a practical framework for researchers who want to maximize what autosomal DNA can do for their genealogy:
Autosomal DNA raw data can be downloaded from all major testing companies and uploaded to FamilyTreeDNA ($19 one-time fee), GEDmatch (free basic / $10 per month Tier 1), and Living DNA (free). MyHeritage discontinued raw DNA uploads in August 2025. AncestryDNA and 23andMe do not accept uploads. GEDmatch’s Lazarus tool requires a Tier 1 subscription.
Sources
- ISOGG Wiki, “Autosomal DNA testing comparison,” isogg.org/wiki/Autosomal_DNA_testing_comparison (accessed May 2026).
- ISOGG Wiki, “Chip SNP overlap comparison,” isogg.org/wiki/Autosomal_DNA_testing_comparison (accessed May 2026).
- GEDmatch, “Lazarus Tool,” gedmatch.com (accessed May 2026).
- GEDmatch, “Lazarus Tool — Reconstructing Deceased Ancestor DNA,” gedmatch.com (accessed May 2026).
- DNA Painter, “About,” dnapainter.com/about (accessed May 2026).
AI Disclosure: This article was researched and written with the assistance of Claude, an AI assistant developed by Anthropic. All facts have been reviewed for accuracy, but readers are encouraged to verify information independently, particularly for time-sensitive details such as pricing and software versions.
Discover more from Dillman Family Association
Subscribe to get the latest posts sent to your email.