Abstract
Statistical tests of association are commonly used to confirm or exclude a relationship between disease and selected genetic variants. Tests based on data from unrelated individuals are popular, but can be invalid in the presence of genetic confounding, for example the sample contains individuals with different genetic ancestries. Family designs that use affected individuals and their relatives can limit the problem of validity using within‐family comparisons. The transmission disequilibrium test (TDT) is the original family test for association proposed for affected offspring and their parents. The test is simple and intuitive, is totally robust and the most powerful against an alternative that assumes an additive mode of inheritance. The family‐based association test (FBAT) is a generalisation of the TDT suitable for other family designs; other modes of inheritance; dichotomous, measured and time‐to‐onset phenotypes and haplotypes of tightly linked markers. FBATs are generally less powerful than tests based on a sample of equivalent size of unrelated individuals, but special settings exist, for example testing for rare variants with affected offspring and their parents, where the family design has a power advantage.
Key Concepts
- Genetic tests based on population or cohort samples can be invalid owing to differing genetic ancestries.
- Association designs using family members can reduce or avoid bias using within‐family comparison.
- The simplest family test is the TDT which uses affected offspring and their parents.
- FBAT refers to a class of tests for family designs that can accommodate any type of genetic model, any phenotype and any family configuration, for example parents and their offspring, sibships or even general pedigrees.
- FBATs are based on the distribution of Mendelian transmissions from parents to offspring and thus only require that Mendel's Laws hold under the null hypothesis of no linkage and no association.
- FBAT tests can accommodate missing parents or founders by conditioning on the observed genotype configuration in the family, or technically, on the sufficient statistics for the missing parents/founders.
- Other methods for testing with family designs that are based on the distribution of genotypes, rather than transmissions, are not completely robust to population substructure in the presence of admixture.
- In general, FBATs are less powerful than their population‐based counterparts, given the same amount of genotyping required, but in some circumstances, for example for rare variants, and in testing for gene–environment interactions, they can be more powerful.
- With rare variants, it is generally considered desirable to consider jointly all rare variants in a specific region. This may require dealing with transmissions of multiple correlated variants that leads to using a number of different approximations.
Keywords: genetic association test; transmission disequilibrium test (TDT); family design; trios; discordant sibships; missing parents; Mendel's laws; Hardy–Weinberg Equilibrium; linkage; linkage disequilibrium; whole‐genome sequencing (WGS); rare variants
