QTDT - Association Models

A model for association may be selected with the -a option. Available choices are:

Option	Description
-af	Fulker Model. Uses siblings to construct a control, and does not require parental genotypes. It should be used in a variance components frame-work.
-ao	Orthogonal Model. This is the Abecasis et al generalization of the Fulker model, allowing for families of any size (with or without parental genotypes). It should be used in a variance components framework. In extended families, a novel scheme for scoring transmission using all available information is used by default (Abecasis 2000b).
-ap	Population Stratification. Evaluate the evidence for population stratification by comparing the between and within family components of association.(This is a variant of the Orthogonal Model)
-aw	Within family component only. Consider only the within family component of association.
-at	Total Association. This evaluates the total evidence for association and is not a TDT. It can produce misleading results in the presence of population stratification.
-ad	Discrete Trait. This is a general pedigree version of the TDT described by Abecasis et al (2000b). The test evaluates whether an allele is transmitted to affected individuals more often than expected by chance.
-aa	Allison's Linear Model [TDTQ5]. This model uses parental mating types to account for stratification. It is appropriate for the analysis of single offspring-parent triads.
-ar	Rabinowitz Model. This is a simple model that produces a sturdy T-test for association. It is also appropriate for families with a single offspring but does not allow for dominance.
-am	Monks Model. This variant of the Rabinowitz model introduces a simple weighing scheme that provides a conservative test of association in families with multiple offspring. (uses parental genotypes if available)
-a-	Do not model association. Useful when evaluating components of variance, for example to estimate an heritability.

A genetic dominance parameter can be included in the analysis by specifying the -g+ option. For example, to specify the Allison model should be used and that dominance should be modelled run the command:

  prompt> qtdt -aa -g+

In linear models, it is often desirable to include covariates that are known to account for some of the variation in the trait. By default QTDT includes all covariates listed in the pedigree file in the analysis, but this behaviour may be modified with the covariate -c option.

Option	Description
-cs	Sex Covariate. Include sex as a covariate.
-cu	User Defined Covariates. Include all user-defined covariates listed in the pedigree and data files.
-cp	Parental Phenotypes. Include parental phenotypes as covariates.

These options may be combined, for example -cps includes both parental phenotypes and sex as covariates, but ignores user-define covariates.