If you have ordered a home paternity test, or you are thinking about ordering one, you have probably wondered what the lab is actually doing with those cheek swabs. The answer starts with something you may have half-remembered from a high school biology class: Gregor Mendel and his pea plants. The rules Mendel worked out in the 1860s are the same rules a modern DNA lab relies on when it reports a 99.99% probability of paternity. This article walks through those rules in plain language, then shows how each one lines up with a step in the paternity testing process. No prior science background needed.
What a gene actually is
A gene is a stretch of DNA that carries instructions for something specific. It might code for a protein that helps your blood clot. It might code for the shape of an enzyme that breaks down lactose. It might do a hundred smaller jobs that add up to a working human body. The National Human Genome Research Institute maintains a plain-language glossary of these terms at the NHGRI Talking Glossary, and it is a useful bookmark if any word here trips you up.
The human genome contains roughly 20,000 protein-coding genes. Every one of them sits at a fixed address on a fixed chromosome. That fixed address matters a lot for what comes next.
Chromosomes and the one-from-each-parent rule
Your DNA is packaged into 46 chromosomes, arranged as 23 pairs. Twenty-two of those pairs are called autosomes and are numbered 1 through 22. The 23rd pair determines biological sex.
Here is the part that matters for paternity testing: one chromosome in each pair came from your biological mother, and the other came from your biological father. Not most of your chromosomes. Not the majority. Exactly one of every pair from each parent, without exception. This split is not random in the sense of being messy or unreliable. It is random in a very controlled, mathematical way, and that controlled randomness is what a DNA test measures.
When a sperm cell forms in a man's body, it carries one chromosome from each of his 23 pairs, chosen essentially at random. When an egg cell forms in a woman's body, the same thing happens on her side. When those two cells combine at conception, the child ends up with 23 chromosomes from mom and 23 from dad, restoring the full set of 23 pairs.
Alleles: the two versions of every gene
Because every gene sits at a fixed address on a chromosome, and because you have two of every chromosome, you also have two copies of almost every gene. Those two copies are called alleles.
Alleles can be identical, or they can be different versions of the same gene. Blood type is a familiar example. The gene that determines your ABO blood type has three common alleles: A, B, and O. You inherit one allele from each parent, so your possible combinations are AA, AO, BB, BO, AB, or OO. That is why two type-A parents can have a type-O child. Each parent quietly carried an O allele alongside their A, and both happened to pass the O.
For paternity testing, labs do not look at blood type genes. They look at short, highly variable regions of DNA called STR markers. But the underlying logic is identical. At every tested location, the child has exactly two alleles, and one came from each biological parent. The Learn.Genetics site from the University of Utah has clear animated walk-throughs of allele inheritance if the concept still feels abstract.
Mendel's two rules, in one paragraph
Mendel described two rules that still hold up today. The first is the law of segregation: when a parent passes DNA to a child, only one of the parent's two alleles for each gene gets passed on, chosen at random. The other allele stays behind. The second is the law of independent assortment: which allele gets passed at one gene has nothing to do with which allele gets passed at another gene on a different chromosome. Each location shuffles on its own. Those two rules are why a child inherits a genuine mix from both parents rather than a carbon copy of one, and they are the mathematical foundation for every paternity test on the market.
How this shows up in DNA paternity testing
A standard home paternity test looks at 23 autosomal STR markers. USDC's home paternity kit can analyze up to 28 genetic markers in total, but paternity itself is answered on the 23-marker standard panel. The extra markers are held in reserve for kinship cases (grandparent, aunt/uncle, sibling) where the standard panel returns an inconclusive Combined Relationship Index and more statistical weight is needed. Industry practice across accredited labs is 20 or more markers. Each marker is a short, repeating stretch of DNA at a known chromosome address, and each one has many possible versions in the general population.
At every one of those 23 marker locations, the lab reads the child's two alleles. It then compares them to the alleged father's two alleles at the same locations. Because of the segregation rule, one of the child's two alleles must match one of the father's two alleles at every single marker. If the man being tested is the biological father, this will be true 23 out of 23 times. If he is not, the pattern will break down quickly.
When a mother's sample is also submitted, the lab can subtract her contribution first. Whichever allele the child has that did not come from mom must have come from the biological father. That leftover allele either matches the tested man or it does not. Including the mother's sample sharpens the result but is not required for a conclusive answer in most cases.
The technical standards for how these markers are chosen, validated, and interpreted are maintained by groups like the International Society for Forensic Genetics, and the reference database of STR markers used across accredited labs lives at the NIST STR DNA Database.
What happens when a marker doesn't match: exclusion
An exclusion is what geneticists call it when the child has an allele at a given marker that could not have come from the tested man. If the child shows a 12 and a 15 at a marker, and the tested man shows a 10 and an 11, there is no way the tested man contributed either of the child's alleles. That is a mismatch.
One mismatch is not automatically a full exclusion, because rare mutations do happen. Labs typically require at least two independent mismatches across the 23 markers before reporting an exclusion. When that threshold is met, the result comes back as 0% probability of paternity, and the report will state that the tested man is excluded as the biological father. The word "excluded" is doing a specific job here: it means the DNA evidence rules him out, not that anyone has judged him or the family situation.
Exclusions are unambiguous. There is no gray area in the math. When the alleles do not match at multiple independent locations, they cannot both have been inherited, and no amount of retesting the same samples will change that.
How the 99.99% number gets built
An inclusion is the opposite of an exclusion. Every marker's alleles are consistent with the tested man being the father. But consistency alone is not proof, because plenty of unrelated men would also happen to carry alleles that fit. What the lab does next is calculate how unlikely that coincidence really is.
At each marker, the lab looks up how common each allele is in the general population. Some alleles are common. Others are rare. Then it calculates a paternity index for that single marker, which is a ratio: how much more likely it is that this specific tested man contributed the observed allele, compared to a random unrelated man from the same population.
Independent assortment is what makes the next step possible. Because the marker results are independent of each other, the lab can multiply the paternity indexes across all 23 markers to get a combined paternity index. That combined number is then converted to a probability of paternity. In a typical inclusion, the combined paternity index runs into the tens or hundreds of thousands, and the probability of paternity comes out at 99.99% or higher.
That 99.99% is not a hedge. It is a statement that the observed DNA pattern is at least 10,000 times more likely if the tested man is the biological father than if a random unrelated man is. A tested man who matches perfectly across 23 highly variable markers is almost impossibly unlikely to be a coincidence. The reason accredited labs stop at 99.99% or 99.9999% and do not report 100% is a statistical convention, not a real uncertainty. You can never fully rule out an identical twin without extra testing, so the ceiling stays just below 100. If you want more on how paternity index calculations work in practice, the NCBI Bookshelf chapter on the human genome covers the underlying population genetics.
What Mendel's rules don't fully explain
Mendel's rules are the workhorse of paternity testing, but real DNA has a few wrinkles the pea-plant experiments did not capture.
The first is recombination. When a parent's cells produce sperm or eggs, chromosomes in each pair physically swap sections with each other before they separate. This means the chromosome you inherited from your mother is not a clean copy of one of her two chromosomes. It is a patchwork of both, stitched together. Recombination is why siblings are not identical even when they inherit the "same" chromosomes at a high level. For paternity testing, recombination is not a problem because the lab is only looking at specific marker locations, and the segregation rule still applies at each one.
The second is mutation. Very rarely, an allele changes during the process of passing from parent to child. A father with alleles 12 and 15 at a given marker might pass on a 13 or a 14 instead. Mutation rates at STR markers are low but not zero, usually well under one percent per marker per generation. This is exactly why labs require multiple mismatches before calling an exclusion. A single odd result at one of 23 markers can be a mutation. Two or three cannot.
The third wrinkle is that some inheritance patterns are more complicated than Mendel described. Sex-linked genes, mitochondrial DNA, and genes that influence each other do not follow simple dominant-recessive rules. None of this affects standard autosomal STR paternity testing, but it is worth knowing that Mendel's rules are the foundation, not the entire building.
Frequently asked questions
Does a child get exactly half their DNA from each parent?
Yes, at the chromosome level. Every child inherits 23 chromosomes from the biological mother and 23 from the biological father. At the individual gene level, the split stays 50-50 for autosomal genes. Sex chromosomes and mitochondrial DNA follow different patterns, but those are not what a paternity test measures.
Can a paternity test be wrong because of a mutation?
Very rarely, a single marker can show a mismatch due to a natural mutation between generations. Accredited labs handle this by requiring at least two independent mismatches across the tested markers before reporting an exclusion. A single anomaly is not enough to change the result.
Why do reports say 99.99% instead of 100%?
Statistical convention. Paternity math is built on comparing your observed DNA pattern to what would be expected from a random unrelated man. Because that comparison always assumes some tiny non-zero chance of coincidence, and because identical twins share DNA, accredited labs do not report absolute 100% figures. A 99.99% result is considered conclusive for personal-knowledge purposes on a home test. If the answer is going to a court, family agency, or immigration filing, a legal chain-of-custody test is required regardless of the probability number.
Does the mother need to be tested for the result to be accurate?
Not usually. Home paternity tests can produce a conclusive result from the child and the alleged father alone. Including the mother's sample makes the calculation more precise and can help resolve ambiguous cases, which is why labs offer the option. USDC's home paternity test kit is $79 and includes the option to add the mother's sample.
What if two brothers might be the father?
This is where standard paternity testing hits its limits. Full brothers share so much DNA that the standard 23-marker panel may not distinguish between them cleanly. In those cases, labs move to specialized kinship testing that examines additional markers and applies different statistical models. USDC's home full sibling DNA test kit is $139 and is designed for exactly these kinds of family relationship questions.
0 comments