calc_score_stats: calc_score_stats.R

In GBJ: Generalized Berk-Jones Test for Set-Based Inference in Genetic Association Studies

calc_score_stats.R

Description

Starting with individual-level data on p factors, generate score test statistics for each factor for input into GBJ/GHC/HC/BJ/minP. Also get the correlations between these test statistics. Designed to be used with linear or logistic or log-linear regression null models.

Usage

calc_score_stats(null_model, factor_matrix, link_function, P_mat = NULL)

Arguments

null_model	An R regression model fitted using glm(). Do not use lm(), even for linear regression!
factor_matrix	An n*p matrix with each factor as one column. There should be no missing data.
link_function	Either "linear" or "logit" or "log"
P_mat	The projection matrix used in calculation may be passed in to speed up the calculation. See paper for details. Default is null.

Value

A list with the elements:

test_stats	The p score test statistics.
cor_mat	The p*p matrix giving the pairwise correlation of every two test statistics.

Examples

Y <- rbinom(n=100, size=1, prob=0.5)
null_mod <- glm(Y~1, family=binomial(link="logit"))
factor_mat <- matrix(data=rnorm(n=100*5), nrow=100)
calc_score_stats(null_mod, factor_mat, "logit")

GBJ documentation built on May 29, 2024, 1:33 a.m.