RNOmni (version 0.2.1)

FIINT: Fully Indirect-INT

Description

Two-stage regression procedure. In the first stage, phenotype is regressed on covariates and population structure to obtain residuals. In the second stage, INT-transformed residuals are regressed on genotype.

Usage

FIINT(y, G, X, S, calcP = T, k = 3/8, parallel = F)

Arguments

y

Numeric phenotype vector.

G

Numeric Snp by obs genotype matrix.

X

Numeric Obs by feature covariate matrix.

S

Numeric Obs by feature structure matrix.

calcP

Logical indicating that p values should be calculated.

k

Offset applied during rank-normalization. See rankNormal.

parallel

Logical indicating whether to run in parallel. Must register parallel backend first.

Value

A numeric matrix of Wald statistics, one for each locus (row) in G, assessing the null hypothesis that genotype is unrelated to the outcome. If p=T, a p-value is additionally calculated for each locus.

Details

Note that, in simulations, the fully indirect approach did not consistently control the type I error. For a similar approach that did provide valid inference, see IINT.

Examples

# NOT RUN {
# FIINT against normal phenotype 
p = RNOmni::FIINT(y=RNOmni::Y[,1],G=RNOmni::G[1:10,],X=RNOmni::X,S=RNOmni::S);
# }