Transmission Disequilibrium Test¶

A family-based statistical test of whether an allele is transmitted from informative heterozygous parents to affected offspring more often than its untransmitted alternative under a specified null.

Version: v1 · 2026-10-17 · History
Domain-specific #: 14340
Domain group: Natural Sciences
Origin domain: Biology & Ecology
Subdomains: Statistical Genetics, Genetic Epidemiology → Biology & Ecology
Aliases: Transmission/disequilibrium test, TDT

Core Idea¶

The transmission disequilibrium test (TDT), in its affected-offspring form, asks whether an allele at a genetic marker is transmitted by heterozygous parents to affected children more often than the parents' alternative allele. Each informative parent supplies the comparison: one of its marker alleles was transmitted and the other was not. The observed directional transmissions are evaluated against a specified no-linkage or equal-transmission null appropriate to the study design. This is a family-based test, not a comparison between unrelated affected people and population controls.^[1]

Spielman, McGinnis and Ewens introduced the test for linkage in the presence of an already observed association near the insulin gene. A later oral-cleft study applied TDT to a different set of affected parent–offspring trios and extended it with conditional-logistic models for effect modification. The extension changes the analysis question without making exposure covariates a required part of the classical test.^[1]^[2]

Structural Signature¶

Affected offspring and family marker genotypes. The child has the selected trait, and parent–child genotypes establish which parental marker allele was passed on. This affected-offspring ascertainment gives the classical test its disease-linkage question. Remove it and an ordinary transmission-distortion study asks a different question.^[1]
Informative heterozygous parent. At the tested marker, the parent has two distinguishable alleles and could transmit either. A homozygous parent has no marker-allele choice for this directional comparison. The original insulin-region analysis identified 57 parents heterozygous for its class-1 versus pooled-class-X contrast.^[1]
Transmitted and untransmitted alternatives. The allele received by an affected child is compared with the alternative in the same parent. This matched control is the distinctive operation; substituting an unrelated person's allele would change the design.^[1]
Transmission-asymmetry test with a declared null. Count or model the directional transmissions and judge their discrepancy against the design's null. The exact statistic and interpretation depend on the allele coding and analytical extension. A pile of transmission counts without a reference expectation is descriptive data, not a completed TDT.^[1]^[2]

What It Is Not¶

TDT is not any association between an allele and disease. An unrelated case-control frequency comparison may show an association, but it lacks the same parent's untransmitted alternative. Nor does an excess transmission by itself prove that the marker allele causes disease: the original question was linkage in the presence of association, and Spielman and colleagues explicitly considered segregation distortion as an alternative explanation for biased inheritance.^[1]

Testing transmission to unaffected or unselected offspring can help probe segregation distortion, but it is not a second positive case of the affected-offspring identity used here. Conditional logistic regression with exposure modifiers is a documented extension, not the defining formula of every TDT.^[1]^[2]

Scope of Application¶

The classical form applies when affected offspring and their parents have usable marker genotypes and at least some parents are heterozygous at the marker under study. It can investigate a candidate region such as the insulin-gene 5-prime flanking polymorphism in families with insulin-dependent diabetes, and it can be used in case-parent trios for a different phenotype, such as nonsyndromic oral clefts. The marker, phenotype, family configuration, allele coding and null must be specified for the claimed result.^[1]^[2]

The original insulin example included families with multiple affected offspring; its reported 124 transmissions are not 124 independent trios. Maestri and colleagues report 160 oral-cleft parent-offspring trios and TDT analyses at markers associated with five candidate genes. Their original abstract also reports covariate effect-modifier analyses, but does not provide the complete marker-by-marker transmission tables or model specifications.^[1]^[2]

Clarity¶

Keep three levels separate: parental genotype, observed transmission, and inference. Heterozygosity says a parent could transmit either marker allele. An observed transmission says which one reached an affected child. A tested excess says the pattern is unlikely under the chosen null at the test's calibration; it does not automatically identify a causal allele or settle every non-disease explanation.^[1]

The untransmitted allele is an internal matched alternative, not an unaffected sibling and not an allele drawn from another population. This distinction explains why a population-frequency difference and a family-transmission difference answer related but nonidentical questions.^[1]

Manages Complexity¶

The test reduces many family genotypes to informative parent-to-affected-child transmission comparisons. The insulin-region analysis pooled marker alleles of classes 2 and 3 as class X, then recorded 78 class-1 transmissions and 46 class-X transmissions from 57 informative 1/X parents to their diabetic offspring. The paper's equal-transmission expectation was 62 versus 62 for those 124 transmissions. That compression makes the directional imbalance visible, while retaining the family design and allele-pooling decisions as conditions of the calculation.^[1]

There is a cost to oversimplification. Parents homozygous at the marker give no directional choice; variant markers and model-based effect-modifier questions cannot be represented merely by relabeling the two cells in the original table. The useful summary is the family-matched contrast plus its declared statistical model, not a universal two-number shortcut.^[1]^[2]

Abstract Reasoning¶

To evaluate a proposed TDT claim, first identify the affected offspring, marker and parental genotypes. Find the informative heterozygous parents and record, for each relevant transmission, which allele was passed on and which alternative remained. State the null and the statistic or model before interpreting an imbalance. In the insulin-region case, Spielman's 78-versus-46 result is compared with a 62-versus-62 expectation; the paper reports a chi-square of 8.26 and p = .004 under its specified calculation.^[1]

Then ask what the imbalance would mean under this sampling design. A within-family contrast helps avoid an artifact from comparing unrelated groups with different allele frequencies, but the inference still depends on ascertainment, genotyping and the genetic/null model. Spielman and colleagues checked unaffected-offspring transmissions because segregation distortion could mimic a disease-linked excess. For Maestri's interaction analysis, the question is more specific: whether a measured covariate changes the modeled transmission effect, not simply whether one pooled count exceeds another.^[1]^[2]

Knowledge Transfer¶

The test can be reused for another affected phenotype or candidate marker when parent–offspring genotypes and the same informative transmission roles are available. The insulin and oral-cleft studies use different phenotypes and markers while preserving the within-parent comparison. Their inferential details do not transfer automatically: a pooled biallelic count in one study does not supply the coefficients or interaction tests in another.^[1]^[2]

Outside statistical genetics, one may recognize the broader idea of a matched alternative tested against a null. That broader idea does not turn a non-genetic matched-pair test into TDT; the named procedure requires parental marker transmission to affected offspring. TDT is a strict child of the live domain-specific Statistical Test genus because it adds those genetic-family roles to a null-calibrated procedure.

Examples¶

Insulin-region marker and insulin-dependent diabetes. In Spielman and colleagues' original analysis, the affected offspring and family marker genotypes were diabetic children and their genotyped parents. The informative heterozygous parents were 57 class-1/X parents at the 5-prime flanking polymorphism. Their transmitted and untransmitted alternatives produced 78 class-1 versus 46 pooled-X transmissions to affected offspring. The declared null comparison expected 62 of each and yielded the paper's reported chi-square result. These counts are transmissions from a multiplex-family sample, not 124 unrelated trios.^[1]

Nonsyndromic oral-cleft case-parent trios. Maestri and colleagues used 160 affected offspring and parent trios and markers associated with five candidate genes. The named TDT entails informative parental transmissions and their untransmitted alternatives; the inspected abstract does not enumerate those count cells. Its test/model includes a TDT analysis and a conditional-logistic extension with covariate effect modifiers. The abstract reports a maternal-smoking interaction for markers near TGFA and TGFB3. That reported interaction is specific to those markers and the study's modeled comparison, not a constitutive role of every TDT.^[2]

Structural Tensions¶

Internal control versus informative-family reach. Conditioning the comparison within parents helps address a population-composition problem of unrelated controls, but the classical contrast needs affected offspring, parental marker data and heterozygous parents. A design can gain a cleaner internal comparison while losing usable information from families lacking an informative parent at the marker. The practical question is how many recruited families actually contribute informative transmissions; merely increasing the number of affected people does not answer it. This is an analytic design trade-off, not a measured yield reported for every TDT study.^[1]

Pooled effect versus effect modification. A pooled transmission excess summarizes one overall contrast. Stratifying or modeling covariates can ask whether that contrast differs across phenotypes or exposures, as in Maestri's oral-cleft study, but it introduces more specific hypotheses and a model-dependent interpretation. Choose the analysis by asking whether the scientific target is overall excess transmission or a specified difference in transmission effect. The original abstract supports the presence of the extension and reported interactions, not a universal power advantage for it.^[2]

Structural–Framed Character¶

On the structural–framed spectrum, TDT is a domain-specific structural method. Its matched transmitted/untransmitted comparison and null are specified operations rather than a loose label for a family resemblance. Its evaluative weight is inferential: a small p value under a model is evidence against that model, not a judgment that an allele is good or necessarily causal.^[1]

Human research practice chooses affected-case ascertainment, marker coding, genotyping and analysis; it does not create the biological inheritance events being measured. Spielman and colleagues specified and named the transmission/disequilibrium procedure for affected-family data.^[1] Words such as “transmission,” “comparison” and “test” travel widely, but importing them into another subject is not recognition of TDT. Recognition requires actual parent–offspring marker transmissions and an affected-offspring null comparison. Its character: a tightly specified genetic-statistical test whose domain-bound roles carry a broader matched-testing skeleton without making the named test cross-domain.

Structural Core vs. Domain Accent¶

The skeletal relation is a matched comparison of two alternatives with observed directional outcomes assessed against a stated null. Here the alternatives are alleles in the same heterozygous parent, the selected outcome is transmission to an affected child, and the inferential target concerns genetic linkage or an explicitly modeled extension. Replace those genetic conditions with generic matched observations and the method may still be a statistical test, but it is no longer this named TDT.^[1]^[2]

The live Statistical Test entry is the approved broader domain-specific genus because TDT supplies data, a null, a test statistic or model, and a declared interpretation. Live Prime Comparison captures the operation of comparing alternatives, but does not itself capture the full matched-alternative, null-calibrated test skeleton. Whether that compound skeleton has independent literal instances across domains and warrants a future Prime is a separate research question. The TDT identity does not clear a substrate-independent Prime bar: every literal case still requires genetic parental transmission to affected offspring.

This entry is a kind of Statistical Test.

The transmission disequilibrium test (TDT) is, in every case, a kind of Statistical Test, and that is its only direct broader abstraction. A statistical test requires observed data, null/reference comparison and declared interpretation; TDT adds affected-offspring ascertainment, informative heterozygous parents and matched transmitted/untransmitted marker alternatives, making it a narrower kind of statistical test. Comparison is a broader general abstraction involved in the procedure, but TDT is not counted as directly falling under it: that relation has not been shown, and TDT's comparing alleles is not enough to establish it.

Population Structure is a related potential confound and motivation for a within-family design. It is not the genus of the test. Linkage disequilibrium is part of the genetic interpretation and title, but it has no entry of its own in the encyclopedia to name as a broader abstraction.

Relationships to Other Abstractions¶

Current abstraction Transmission Disequilibrium Test Domain-specific

Parents (1) — more general patterns this builds on

Transmission Disequilibrium Test is a kind of Statistical Test Domain-specific

TDT is a null-calibrated statistical test specialized to informative parental marker transmissions to affected offspring.

Hierarchy path (1) — routes to 1 parentless root

Transmission Disequilibrium Test → Statistical Test

Neighborhood in Abstraction Space¶

Transmission Disequilibrium Test sits in a sparse region of the domain-specific corpus (96^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (2721 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-17

Not to Be Confused With¶

Unrelated case-control association: compares allele frequencies in separately sampled groups, not the transmitted and untransmitted alternatives within the same heterozygous parent.^[1]
Segregation-distortion test using unselected offspring: can reveal biased inheritance, but lacks the affected-offspring disease-linkage sampling question used here. Spielman's unaffected-offspring comparison served as an alternative-explanation check.^[1]
An allele's causal effect: a TDT signal is a statistical result under stated assumptions; it does not alone establish that the tested marker is causal.^[1]
The conditional-logistic extension: Maestri used covariates as effect modifiers, including a maternal-smoking interaction for particular markers; these are additional modeled questions, not universal components of the classical test.^[2]

References¶

[1] Spielman, Richard S., Ralph E. McGinnis, and Warren J. Ewens (1993), "Transmission Test for Linkage Disequilibrium, The Insulin Gene Region and Insulin-Dependent Diabetes Mellitus (IDDM)" (printed title uses a colon after “Disequilibrium”). American Journal of Human Genetics 52, 506–516. Original full article inspected, especially pp. 507–512, equation (5), and tables 5–6. https://pmc.ncbi.nlm.nih.gov/articles/PMC1682161/ registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o ↩p ↩q ↩r ↩s ↩t ↩u ↩v ↩w ↩x ↩y

[2] Maestri, N. E., et al. (1997), "Application of Transmission Disequilibrium Tests to Nonsyndromic Oral Clefts, Including Candidate Genes and Environmental Exposures in the Models" (printed title uses a colon after “Clefts”). American Journal of Medical Genetics 73(3), 337–344. Original indexed abstract and bibliographic metadata inspected; full article not inspected. https://pubmed.ncbi.nlm.nih.gov/9415696/ registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l