getG0: Obtain the consensus null

In CenterForStatistics-UGent/reconsi: Resampling Collapsed Null Distributions for Simultaneous Inference

Obtain the consensus null

Description

Obtain the consensus null

Usage

getG0(
  statObs,
  statsPerm,
  z0Quant,
  gridsize,
  maxIter,
  tol,
  estP0args,
  testPargs,
  B,
  p,
  pi0,
  assumeNormal,
  resamAssumeNormal
)

Arguments

statObs	A vector of lenght p with observed test statistics
statsPerm	A pxB matrix with permuation z-values
z0Quant	a vector of length of quantiles defining the central part R of the distribution. If a single number is supplied, then (z0quant, 1-z0quant) will be used
gridsize	An integer, the gridsize for the density estimation
maxIter	An integer, the maximum number of iterations in determining R
tol	The convergence tolerance.
estP0args	A list of arguments passed on to the estP0args() function
testPargs	A list of arguments passed on to quantileFun
B	an integer, the number of permutations
p	an integer, the number of hypotheses
pi0	A known fraction of true null hypotheses
assumeNormal	A boolean, should normality be assumed for the null distribution?
resamAssumeNormal	A boolean, should normality be assumed for resampling dists

Value

A list with following entries

PermDensFits	The permutation density fits
zSeq	The support of the kernel for density estimation
zValsDensObs	The estimated densities of the observed z-values
convergence	A boolean, has the algorithm converged?
weights	Vector of length B with weights for the permutation distributions
fdr	Estimated local false discovery rate along the support of the kernel
p0	The estimated fraction of true null hypotheses
iter	The number of iterations
fitAll	The consensus null fit

CenterForStatistics-UGent/reconsi documentation built on Nov. 13, 2023, 2:04 a.m.