BOOST: Implements BOOST SNP-SNP interaction test

In epiGWAS: Robust Methods for Epistasis Detection

Description

For a pair of SNPs (X1, X2) and a binary phenotype Y, the BOOST function computes the ratio of maximum log-likelihoods for two models: the full model and the main effects model. Mathematically speaking, the full model is a logistic regression model with both main effects and interaction terms (X1, X2, X1, X1 x X2). The main effects model is a logistic regression model with only (X1, X2) as covariates. Since we are interested in the synergies with a single variant, we do not implement the initial sure screening stage in BOOST which filters out non-significant pairs.

Usage

BOOST(A, X, Y, ncores = 1)

Arguments

A	target variant. The SNP A is encoded as 0, 1, 2.
X	genotype matrix (excluding A). The only accepted SNP values are also 0, 1 and 2.
Y	observed phenotype. Binary or two-level factor.
ncores	number of threads (default 1)

Value

The interaction statistic between each column in X and A

See Also

The webpage http://bioinformatics.ust.hk/BOOST.html provides additional details about the BOOST software

Examples

X <- matrix((runif(500, min = 0, max = 1) < 0.5) +
    (runif(500, min = 0, max = 1) < 0.5), nrow = 50)
A <- (runif(50, min = 0, max = 1) < 0.5) + (runif(50, min = 0, max = 1) < 0.5)
Y <- runif(50, min = 0, max = 1) < 1/(1+exp(-.5 * A * X[, 3] + .25 * A * X[, 7]))
BOOST(A, X, Y)

epiGWAS documentation built on Sept. 8, 2019, 5:02 p.m.