Zachary R. McCaw
Updated: 2026-02-24
Genetic association tests based on the inverse normal transformation (INT). Recommended for continuous traits with non-normally distributed residuals. INT-based tests control type I error when standard linear regression does not (e.g. skewed or kurtotic residuals) and typically outperform linear regression in power.
- D-INT (direct): the phenotype is rank-normalized, then tested.
- I-INT (indirect): phenotypic residuals are rank-normalized, then tested.
- O-INT (omnibus): combines D-INT and I-INT via Cauchy combination for a single robust test.
Reference: McCaw et al., Biometrics (2020).
# CRAN.
install.packages("RNOmni")
# GitHub.
remotes::install_github("zrmacc/RNOmni", build_vignettes = TRUE)| Function | Description |
|---|---|
OINT() |
Omnibus INT test |
DINT() |
Direct INT test |
IINT() |
Indirect INT test |
BAT() |
Basic association test (no transformation) |
RankNorm() |
Rank-based inverse normal transform |
OmniP() |
Cauchy combination of p-values |
library(RNOmni)
set.seed(100)
n <- 500
X <- cbind(1, rnorm(n))
G <- replicate(100, rbinom(n, 2, 0.25))
storage.mode(G) <- "numeric"
y <- exp(X %*% c(1, 0.5) + rnorm(n)) # skewed phenotype
p <- OINT(y = y, G = G, X = X, simple = TRUE)