Informative Meioses and Their Recombination Evidence

Why Informative Meioses Matter

Linkage analysis asks whether alleles at two loci tend to be transmitted together more often than would be expected if the loci assorted independently. In a family, the relevant evidence comes from parental meioses: which alleles did a parent transmit to a child, and do those alleles match the parent’s phased haplotypes? The offspring’s genotype does not always reveal the transmitted haplotype. Recognizing that distinction is essential: a meiosis whose recombinant status cannot be resolved should not be counted as either a recombinant or a nonrecombinant.

This tutorial considers a disease locus with alleles \(D\) and \(d\), and a marker locus with alleles \(M\) and \(m\). The letters are labels, not claims about dominance or disease status. We assume the parent’s phase is known, the parental and offspring genotypes are observed without error, and there is no mutation between parent and child. The goal is to classify a transmission from those genotypes, not to infer disease status from phenotype.

Phase Determines What Counts as Recombinant

A parent heterozygous at both loci has two possible phases. In coupling phase, the alleles \(D\) and \(M\) are on one chromosome, and \(d\) and \(m\) are on the other: \(D\!-\!M/d\!-\!m\). In repulsion phase, the haplotypes are \(D\!-\!m/d\!-\!M\). Phase is the arrangement of alleles across homologous chromosomes; the two unphased genotypes alone do not tell us which arrangement is present.

For a parent with known phase, a nonrecombinant transmission carries one of that parent’s original haplotypes. A recombinant transmission carries a combination of alleles that were on opposite parental haplotypes. Thus, the same transmitted allele pair can be recombinant in one phase and nonrecombinant in the other.

Let \(\theta\) be the recombination fraction between the two loci, with \(0\leq\theta\leq 1/2\). Conditional on the transmitted disease-locus allele and the known parental phase, the probability that the marker allele came from the same parental haplotype is \(1-\theta\); the probability it came from the other haplotype is \(\theta\). These probabilities are useful only when the observations distinguish the two possibilities. The condition \( \theta\leq 1/2\) reflects the usual two-locus linkage parameterization: a recombination fraction above one half does not provide a distinct two-point linkage interpretation.

When the Meiosis Is Uninformative

A meiosis is uninformative when the observed genotypes do not reveal whether the transmitted marker allele was on the same parental haplotype as the transmitted disease-locus allele. A common case is that the parent is homozygous at the marker. If both parental marker alleles are \(M\), the child necessarily receives \(M\) from that parent, regardless of which disease-locus allele was transmitted and regardless of whether recombination occurred. The marker genotype then cannot distinguish the two recombination statuses.

The likelihood calculation makes the point precise. When an observation is compatible with either status, its probability is obtained by summing over those unobserved possibilities, weighted by their probabilities. If the observation is equally compatible with one nonrecombinant state and one recombinant state, the conditional contribution is

This contribution does not vary with \(\theta\). Such a meiosis gives no direct evidence favoring one recombination fraction over another. It is not a recombinant transmission with uncertain status, nor is it a nonrecombinant transmission; its status is unresolved.

More generally, it is the probability of the observed genotype after summing over all compatible transmission states that matters. If those states do not yield a probability that changes with \(\theta\), the observation contributes no information about \(\theta\) under the stated model. If the data only partially resolve the transmission, the contribution need not be exactly one or entirely uninformative; it must be calculated by summing the probabilities of the compatible states. Missing genotypes, ambiguous phase, or genotype uncertainty may therefore require more than a simple binary classification.

Why Uninformative Meioses Do Not Count as Either Outcome

For an informative meiosis, the observed genotypes single out a transmission category. Relative to independent assortment, which corresponds to \(\theta=1/2\), a resolved nonrecombinant transmission has conditional likelihood ratio

and a resolved recombinant transmission has conditional likelihood ratio

For instance, at \(\theta=0.10\), these are \(1.8\) and \(0.2\), respectively. These ratios express how the observed, resolved transmission compares under a specified recombination fraction with the value at independent assortment. By contrast, an unresolved observation whose probability is \((1-\theta)+\theta=1\) has ratio \(1/1=1\): it does not change the comparison. This is why simply counting all children or all parent-to-child transmissions can misrepresent how much linkage evidence a pedigree contains.

The equality above also gives a short justification for treating some ambiguous meioses as neutral. If the data cannot distinguish one nonrecombinant possibility from one recombinant possibility, then the two probabilities partition the possibilities: \(1-\theta\) for the first and \(\theta\) for the second. Their sum is one for every allowed \(\theta\). Since the observed genotype has the same probability across the linkage parameter values, it cannot favor one value over another in this conditional comparison.

In a pedigree, the contribution of each meiosis depends on what can be inferred from its own parental phase and observed offspring genotypes. Resolved transmissions can provide evidence in either direction; unresolved ones may provide none. When combining evidence across meioses, the likelihood accounts for the probability of the observed data, including uncertainty about transmission state, rather than assigning an unsupported recombinant/nonrecombinant label.

A Practical Classification Workflow

A careful classification can be made in a consistent order. This process is especially helpful when several parents or children appear in the same pedigree.

Common Mistakes and Why the Distinction Matters

• Ignoring phase. The allele pair \(D\!-\!M\) is nonrecombinant in coupling and recombinant in repulsion. Always compare the transmission with the actual parental haplotypes.
• Calling every heterozygous parent informative. Heterozygosity at both loci makes phase and transmission potentially useful, but the offspring genotypes must still identify the transmitted alleles. Otherwise, the meiosis may remain unresolved.
• Counting an ambiguous transmission as a recombinant. Uncertainty is not evidence of recombination. Sum the probabilities of compatible states; in the simple one-of-each case, \((1-\theta)+\theta=1\).
• Interpreting a nonrecombinant transmission as proof of no crossover. A nonrecombinant observed haplotype is a classification of the alleles transmitted. It does not mean a crossover could not have occurred elsewhere or that the model’s recombination probability was zero.
• Confusing a transmission probability with a conclusion about linkage. A single classified meiosis is one piece of evidence. Linkage conclusions depend on the likelihood across the observed pedigree data and the model assumptions.

This distinction matters because a pedigree’s size alone does not determine how much information it contains. Several children may have genotypes that leave parental transmissions unresolved, while a smaller number of well-resolved meioses may be informative. Linkage evidence comes from what the observations distinguish under the phased inheritance model, not merely from the number of family members recorded.