simico_out: Get P-Values

In SIMICO: Set-Based Inference for Multiple Interval-Censored Outcomes

Get P-Values

Description

Calculate test statistic and p-values for multiple outcome test and multiple burden test.

Usage

simico_out(nullFit, xDats, lt_all, rt_all, Itt, a1, a2, G, k, d)

Arguments

nullFit	Results of the Newton-Raphson: estimates of the beta coefficients.
xDats	List of design matrices.
lt_all	Matrix containing the generated left interval times.
rt_all	Matrix containing the generated right interval times.
Itt	I_theta theta - Information matrix of theta.
G	n x q matrix of genetic information.
a1	First shape parameter of beta parameter.
a2	Second shape parameter of beta parameter.
k	Total number of outcomes.
d	Number of quadrature nodes.

Value

multQ	Score statistic for multiple outcome test.
multP	P-value for multiple outcome test.
burdQ	Score statistic for multiple burden test.
burdP	P-value for multiple burden test.

Examples

# Set number of outcomes
k = 2

# Set number of observations
n = 100

# Set number of covariates
p = 2

# Set number of SNPs
q = 50

# Set number of causal SNPs
num = 5

# Set number of quadrature nodes
d = 100

# Variance of subject-specific random effect
tauSq = 1

# Define the effect sizes
effectSizes <- c(.03, .15)

# Set MAF cutoff for causal SNPs
Causal.MAF.Cutoff = 0.1

# the baseline cumulative hazard function
bhFunInv <- function(x) {x}

set.seed(1)

# Generate covariate matrix
xMat <- cbind(rnorm(n), rbinom(n=n, size=1, prob=0.5))

# Generate genetic matrix
gMat <- matrix(data=rbinom(n=n*q, size=2, prob=0.1), nrow=n)

# Get indices to specific select causal variants
idx <- Get_CausalSNPs_bynum(gMat, num, Causal.MAF.Cutoff)

# Subset the gMat
gMatCausal <- gMat[,idx]

# Generate the multiple outcomes
exampleDat <- simico_gen_dat(bhFunInv = bhFunInv, obsTimes = 1:3,
                             windowHalf = 0.1, n, p, k, tauSq, gMatCausal,
                             xMat, effectSizes)

# Set the initial estimate values
init_beta <-c (rep(c(0, 0, 0, 1, 0), k), 1)

# Run the newton-raphson
nullFit <- simico_fit_null(init_beta = init_beta,
   epsilon = 10^-5, xDats = exampleDat$fullDat$xDats,
   lt_all = exampleDat$leftTimesMat,
   rt_all = exampleDat$rightTimesMat,
   k = k, d = d)

# Get the test statistics p-values
out <- simico_out(nullFit = nullFit$beta_fit,
   xDats = exampleDat$fullDat$xDats,
   lt_all = exampleDat$leftTimesMat,
   rt_all = exampleDat$rightTimesMat,
   Itt = nullFit$jmat, a1 = 1, a2 = 25,
   G = gMat, k  = k, d = d)

# Print results
# Score statistic
out$multQ

# P-values
out$multP

SIMICO documentation built on Nov. 26, 2022, 1:06 a.m.