Why Founder Assumptions Matter
A pedigree traces relationships among the people included in a family diagram. It rarely displays every ancestor. At some point, a person's parents are not shown, and the diagram stops. Calculations involving that person then need a rule for the unshown ancestry. Treating the person as a founder with unrelated ancestors is one possible rule, but it is an assumption, not a fact established by the drawing.
This distinction matters because pedigree calculations have two parts. Mendelian transmission describes how alleles pass from known parents to children. Founder assumptions describe the alleles assigned to people at the pedigree boundary, and whether those people share ancestry with one another. The transmission rule can be correct while the assumptions about the boundary are inappropriate.
What Is Assumed About Founder Alleles?
At a particular genetic locus, a person has two allele copies. A founder model specifies how those copies are distributed before the observed pedigree's transmissions are considered. For example, at a biallelic locus with alleles \(A\) and \(a\), let their population frequencies be \(p\) and \(q=1-p\). If the founder's two copies are modeled as independent draws from this population, the genotype probabilities are \(P(AA)=p^2\), \(P(Aa)=2pq\), and \(P(aa)=q^2\). This is the Hardy–Weinberg genotype model.
That model uses assumptions: allele frequencies are specified for the relevant population; the two copies are sampled independently; and the locus is modeled without mutation in the pedigree transmissions. The population model may also assume random mating. These are modeling choices, not consequences of being labeled a founder.
A second assumption concerns different founders. If two founders are treated as unrelated draws from a population with fixed allele frequencies, their genotypes can be modeled as independent. “Unrelated” here means that the model does not include shared ancestry connecting them within the pedigree being analyzed. It does not mean that they must carry different allele types: two unrelated people can both carry \(A\), by chance.
Unrelatedness and the allele-frequency model are distinct. A population-frequency model specifies the distribution of a founder's genotype. An unrelatedness assumption specifies whether separate founders' genotypes are independent. If founders share ancestry, multiplying their separate genotype probabilities may be wrong, even if each person's marginal genotype probabilities are known.
When Founder Probabilities Factor
Suppose \(G_1\) and \(G_2\) are the genotypes of two founders. If the model says they are independent conditional on the specified population allele frequencies, then for any particular genotypes \(g_1\) and \(g_2\),
The justification is the definition of independence: the joint probability of the two events equals the product of their probabilities. If the founders are related, or if the model includes shared unobserved ancestry or uncertain population frequencies that induce dependence, that factorization need not hold. A pedigree diagram with two separate branches does not establish independence; the branches may simply omit a shared ancestor.
Worked Example: Genotype Probabilities for Two Unrelated Founders
Consider a hypothetical autosomal locus with alleles \(A\) and \(a\), with \(p=P(A)=0.60\) and \(q=P(a)=0.40\). Assume each founder's two allele copies are independent population draws, and assume two founders are unrelated draws from this same population with fixed allele frequencies. Find the probability that both founders are heterozygous.
For one founder, the Hardy–Weinberg probability of genotype \(Aa\) is
Under the stated unrelatedness assumption, the two founder genotypes are independent. Therefore,
The model assigns probability 0.2304 to both founders being heterozygous. This result depends on the stated allele-frequency and independence assumptions; the pedigree labels alone do not supply them.
What a Pedigree Boundary Does—and Does Not—Tell You
When a person's parents are absent from a diagram, their genotypes and ancestry are not observed there. The person is a founder for that diagram, but may be a descendant of relatives, may share ancestors with another displayed founder, or may come from a population whose allele frequencies differ from those assumed. A boundary therefore marks missing information. It does not prove that the omitted history is irrelevant.
For a child whose parents' genotypes are known, Mendelian transmission can still be calculated without knowing the grandparents. For example, if one parent is \(Aa\) and the other is \(aa\), the heterozygous parent transmits \(A\) or \(a\) with probability \(1/2\), and the \(aa\) parent transmits \(a\) with probability 1. The ancestry omitted above those parents does not change this conditional transmission calculation. It can, however, matter when the parental genotypes themselves are unknown and must be assigned probabilities.
This is a useful boundary check: ask whether the calculation is conditional on known genotypes, or whether it requires a probability model for those genotypes. In the first case, apply the transmission rules to the known parental genotypes. In the second, state how founder genotypes are distributed and whether the relevant founders are modeled as independent.
Worked Example: A Transmission Probability Despite Missing Grandparents
In a hypothetical pedigree, a child's mother is known to have genotype \(Aa\), and the father is known to have genotype \(aa\). Neither parent's parents appear in the diagram. Find the probability that the child has genotype \(Aa\), conditional on these parental genotypes.
The mother transmits \(A\) with probability \(1/2\) and \(a\) with probability \(1/2\). The father transmits \(a\) with probability 1. Thus the child is \(Aa\) exactly when the mother transmits \(A\):
The answer is 0.5 conditional on the stated parental genotypes and ordinary Mendelian transmission. The missing grandparents do not prevent this calculation. They would matter if the task instead required deriving probabilities for the mother's or father's genotype from their ancestry.
Recognizing a Missing-Ancestry Problem
A pedigree boundary becomes important when two branches are treated as independent even though the diagram does not establish that they are. For instance, two people may each have unshown parents, yet those parents could be related. If the relationship is known, the pedigree should include the shared ancestry needed for the question, or the analysis should state an explicit model for that relationship. If the relationship is unknown, the available diagram alone cannot determine whether independent founder probabilities are justified.
The following logic makes the issue precise. Separate marginal probabilities \(P(G_1=g_1)\) and \(P(G_2=g_2)\) do not by themselves determine the joint probability \(P(G_1=g_1,G_2=g_2)\). Multiplying the marginals gives the joint probability only under independence. Thus, missing ancestry can make a joint genotype calculation underdetermined, even when each founder's marginal genotype distribution is specified.
Worked Example: Why Two Separate Branches May Not Be Independent
Suppose a hypothetical analysis assigns each of two boundary individuals a marginal probability of 0.30 of genotype \(AA\). A diagram shows no connection between them because their parents are omitted. An analyst multiplies the marginals and reports a probability of \(0.30\times0.30=0.09\) that both are \(AA\).
That multiplication is valid only if the model treats the two genotypes as independent. To see why the marginals alone are insufficient, suppose an explicitly specified joint model instead gives \(P(G_1=AA,G_2=AA)=0.12\). This joint value is compatible with both marginals being 0.30: the probabilities of \(G_1=AA,G_2\ne AA\) and \(G_1\ne AA,G_2=AA\) can each be 0.18, leaving 0.52 for neither being \(AA\). These four probabilities sum to 1 and give each marginal \(0.12+0.18=0.30\).
Under that joint model, the probability that both are \(AA\) is 0.12, not 0.09. The value 0.12 is not inferred from the incomplete pedigree; it is included to demonstrate that the marginals do not fix the joint probability. The analyst must justify independence or obtain a joint model informed by the omitted ancestry.
A Practical Boundary Checklist
Before calculating from a pedigree, make the assumptions visible. This is especially important when the diagram is a partial family history or when founder genotypes are not observed.
Identify every person whose parents are not shown. Treat “founder” as a role in this displayed pedigree, not as a claim about complete ancestry.
If parental genotypes are known, calculate transmission conditional on them. If they are unknown, specify the genotype model used to assign them probabilities.
Use a product of founder probabilities only when the model supports independence. Separate branches in a drawing are not enough.
If omitted ancestry could connect the branches, say that the joint calculation cannot be determined from the displayed pedigree alone unless an additional relationship model is supplied.
Common Mistakes
- Equating “founder” with “unrelated.” A founder is simply someone whose parents are not shown. Unrelatedness must be assumed or supported by information beyond that label.
- Assuming unrelated people carry different alleles. Unrelated founders can carry the same allele type by chance. Unrelatedness concerns the model for their ancestry, not whether their allele labels match.
- Multiplying marginal probabilities automatically. The product \(P(G_1=g_1)P(G_2=g_2)\) is the joint probability only under independence.
- Blaming a transmission calculation on missing grandparents. If both parental genotypes are known, a child's conditional transmission probability can often be calculated directly. Missing ancestry matters when it is needed to model those genotypes or their dependence.
- Concealing a boundary assumption. If the pedigree is incomplete, state whether the analysis assumes unrelated founders, uses known shared ancestry, or cannot determine the required joint distribution.
Check Your Understanding
Answer each question using the assumptions stated in the prompt.
- In a displayed pedigree, neither parent of person X is shown. What does “founder” mean for X, and what does it not establish about X's ancestry?
- At a biallelic locus, \(p=0.70\) and \(q=0.30\). Under the independent-allele Hardy–Weinberg model, calculate the probability that one founder is heterozygous.
- Two founders each have a specified marginal probability of genotype \(Aa\). What additional condition permits multiplying those probabilities to find the probability that both are \(Aa\)?
- A child's two parental genotypes are known, but the grandparents are not shown. Explain whether the missing grandparents prevent a conditional Mendelian transmission calculation.
- Two pedigree branches stop at different people, but their omitted ancestors might be shared. What can and cannot be concluded about independence from the diagram alone?