mztwinreg (version 1.0-1)

fam_env: Linear and logistic regression models for familiar and unique environmental factors

Description

This function implements linear and logistic regression models to test for the association between an outcome phenotype and both the familial (genes + shared environment) and the unique environmental influences on a predictor variable.

Usage

fam_env(formula, BbBw = NULL, regression = "linear", cluster = "default",  adjust = "robcov", robcov_method = "huber", bootcov_B = 200, data, ...)

Arguments

formula

an object of class formula (or one that can be coerced to that class): a symbolic description of the model to be fitted. See formula.

BbBw

the name of the predictor variable(s) (already in formula) for which the familial and the unique environmental influences are going to be evaluated.

regression

the type of regression model to be fitted. Use either 'linear' (default) or 'logistic'.

cluster

a vector identifying the pairs in the sample. If not specified, the default value assumes that twin pairs are introduced in adjacent rows.

adjust

a method to adjust for correlated responses (heteroskedasticity) of twin pairs. Use either 'robcov' (default) or 'bootcov'.

robcov_method

if adjust = 'robcov' (default) is selected, it allows choosing a method to adjust the variance-covariance matrix. Use either 'huber' (default: Huber-White sandwich estimator) or 'efron' (especially for small samples; see robcov).

bootcov_B

if adjust = 'bootcov' is selected, it allows specifying the number of bootstrap repetitions to computes an estimate of the covariance matrix for a set of regression coefficients.

data

a data frame containing the variables in the model.

...

additional arguments to be passed to either ols (linear regression) or lrm (logistic model), from the rms package.

Value

fam_env returns an object of class "rms", along with either c("ols", "lm") (linear models) or c("lrm", "glm") (logistic models).

Details

As shown by Carlin, J. B. et al. (1994), these cluster-based regression models allow parsing out familial and environmental factors contributing to the value of a predictor variable.

References

Carlin, J. B., Gurrin, L. C., Sterne, J. A., Morley, R., & Dwyer, T. (2005). Regression models for twin studies: a critical review. International Journal of Epidemiology, 34(5), 1089-1099.

See Also

rms, ols, lrm, robcov, bootcov

Examples

data(flu_weight)

# The linear regression below tests whether the intrapair differences in 
# DNA methylation of a given twin-pair are predicted by its differences in
# either familial (Bb) or environmental (Bw) factors influencing weight 
# (in kilograms). 
(fam_env_linear <- fam_env(DNAmeth ~ Gender + Age + Weight, BbBw="Weight", 
regression='linear', data=flu_weight))

# The linear regression below tests whether the intrapair differences in
# flu (binary outcome) of a given twin-pair are predicted by its differences
# in either familial (Bb) or environmental (Bw) factors influencing weight 
# (in kilograms). 
(fam_env_logistic <- fam_env(FluNow ~ Gender + Age + Weight, BbBw="Weight", 
regression='logistic', data=flu_weight))