expit: The expit function

In tamartsi/BinomiRare: Test the association of rare variants with a disease

Description

returns the expit of a variable x expit(x) = exp(x)/(1 + exp(x))

Usage

expit(x)

Arguments

x	a numerical vector

Details

applies the expit transformation

Value

expit(x)

Note

Useful for calculating disease probabilities from logistic regression.

Author(s)

Tamar Sofer

Examples

n <- 10000
effect.size <- 1
pop.risk <- -2.6

x <- rnorm(n, sd = 0.01)
x <- pmax(x, 0)
	
g <- rbinom(n, size = 2, prob = x)  ## one causal variant, x is a confounder
G <- matrix(rbinom(n*100, size = 1, prob = 0.001), nrow = n)  ### another 100 null variants
G <- cbind(g, G)
colnames(G) <- paste0("var_", 1:ncol(G))
rownames(G) <- paste0("person_", 1:nrow(G))
p <- expit(pop.risk +   g*effect.size + 20*x)
d <- rbinom(n, 1, p) 
names(d) <- paste0("person_", 1:nrow(G))


########### Now that we have outcome d, genotypes G and a covariate x: 
########### Estimate disease probability model

prob.mod <- glm(d ~ x, family = binomial)
prob.d <- expit(predict(prob.mod))

tamartsi/BinomiRare documentation built on May 31, 2019, 2:56 a.m.