familyBasedCITest: Conditional Independence Test for Family Data

In adele/FamilyBasedPGMs: Methods for Learning Genetic and Environmental Graphical Models from Family Data

Description

Computes the p-value of the test of the null hypothesis of zero genetic, environmental, or total partial correlation between X and Y given S, if dagg, dage, or dagt is set as TRUE, respectively.

The unconfounded estimation and significance test for the genetic and environmental partial correlation coefficients are described in detail in \insertCiteribeiro2019family;textualFamilyBasedPGMs. They are based on univariate polygenic linear mixed models \insertCitealmasy1998multipointFamilyBasedPGMs, with two components of variance: the polygenic or family-specific random effect, which models the phenotypic variability across the families, and the environmental or subject-specific error, which models phenotypic variability after removing the familial aggregation effect.

This function can be used as a d-separation oracle in causal structure learning algorithms such as the IC/PC algorithm when the variables are normal distributed and the observations are clusterized in families.

Usage

familyBasedCITest(x, y, S, suffStat)

Arguments

x	An integer value indicating the position of variable X.
y	An integer value indicating the position of variable Y.
S	An integer vector with the positions of zero or more conditioning variables in S.
suffStat	A list with the elements "phen.df", "covs.df", "pedigrees", "minK", "maxFC", "orthogonal", "alpha", "dirToSave", "fileID", "savePlots", and "useGPU", as described in function learnFamilyBasedDAGs, "dagg", "dage", "dagt", to specify whether the genetic, environmental, or total conditional independence test should be conducted, and also "Phi2" and "Z", both obtained by the function calculateBlockDiagonal2PhiMatrix with the argument "squaredRoot" set, respectively, as FALSE and TRUE.

Value

The p-value of the test.

References

\insertAllCited

adele/FamilyBasedPGMs documentation built on Feb. 16, 2021, 8:29 a.m.