Skip to main content

Table 1 Regression models included in the paper. The 'Model' column refers to the number that identifies each model in the paper.

From: Gene-environment interaction tests for family studies with quantitative phenotypes: A review and extension to longitudinal measures

Models for cross-sectional data

  

Model

Independent subjects design

Case-parent design

Comments

Main genetic effect

   

(2)

E(Y i |X, Z) = β0 + β1X i + β2Z i

 

Ordinary linear regression model

(3)

 

E(Y i |X, Z) = β0 + β1(X i - E(X i |g im , g if )) + β2Z i

Adjusted version of (2). The model is adjusted by the expected value of the offspring's genotype conditional to the parental genotypes

(4)

 

E(Y i |X, Z) = β0M+ β1(X i - E(X i |g im , g if )) + β2Z i

Gauderman's model (QTDTM) adjusted for the covariate Z

(5)

 

E(Y i |X, Z) = β0M+ β1X i + β2Z i

(4) equivalent to (5)

(10)

 

χ F B A T 2 = U 2 V a r U

FBAT statistic

Gene-environment interaction

  

(1)

E(Y i |X, Z) = β0 + β1X i + β2Z i + β3X i Z i

 

Ordinary linear regression model

(6)

 

E(Y i |X, Z) = β0M+ β1X i + β2Z i + β3X i Z i

Gauderman's model (QTDTM)

(7)

 

E(Y i |X, Z) = β0M+ β1[X i - E(X i |g im , g if )] + β2Z i + β3Z i [X i - E(X i |g im , g if )]

(6) is not equivalent to (7) when the environment covariate (Z i ) is not constant within mating type.

(8)

 

E(Y i |X, Z) = β0M+ β1[X i - E(X i |g im , g if )] + β2MZ i + β3Z i [X i - E(X i |g im , g if )]

Adjusted QTDTM

(9)

 

E(Y i |X, Z) = β0M+ β1X i + β2MZ i + β3Z i X i

(8) equivalent to (9)

Main genetic effect

  

(19)

E(Y ij |X, Y, t) = α0 + α1t ij + α2Z i + α3X i + α4X i Z i + α5X i t ij + α6Z i t ij

 

Ordinary linear mixed model (OLMM)

(20)

 

E(Y ij |X, Y, t) = α0M+ α1Mt ij + α2Z i + α3X i + α4X i Z i + α5X i t ij + α6MZ i t ij

Adjusted linear mixed model (ALMM)

Gene-environment interaction

  

(11)

Y ij = α0 + α1t ij + α2Z i + α3X i + α4X i Z i + α5X i t ij + α6Z i t ij + α7X i Z i t ij + b1i+b2it ij + e ij

 

Ordinary linear mixed model

(12)

 

FEF2575ij= α0M+ α1Mt ij + α2Z i + α3[X i - E(X i |g im , g if )] + α4[X i - E(X i |g im , g if )]Z i + α5[X i - E(X i |g im , g if )]t ij + α6MZ i t ij + α7[X i - E(X i |g im , g if )]Z i t ij + b1i+b2it ij + e ij

Adjusted linear mixed model (ALMM)

(13)

 

FEF2575ij= α0M+ α1Mt ij + α2Z i + α3X i + α4X i Z i + α5X i t ij + α6MZ i t ij + α7X i Z i t ij + b1i+b2it ij + e ij

(13) is equivalent to (12)

  1. X i is a fixed variable that translates an offspring genotype to a numerical value; Z i is an observed environmental covariate, either continuous or dichotomous; g im , g if are the parental genotypes (mother and father, respectively); E(X i |g im , g if ) is calculated under segregation and independent assortment assumptions using Mendel's law; M = 1, 2, ..., 6 are the six possible mating types; i = 1, 2, 3, ..., n subjects; j = 1, 2, 3, ..., m measurement occasions into the subject; t ij is the repeated time (or exposure) variable;
  2. b1iis the random subject intercept effect; (α0 + b1i) varies among subjects; b2iis the random subject slope effect: (α1 + b2i)t ij varies among subjects; e ij is a random variable regarded as measurement or sampling errors.