CBUM: (Censored) Beta-Uniform mixture for p-values

In gitlongor/pi0: Estimating the Proportion of True Null Hypotheses for FDR

Description

This function implements the method of Markitsis and Lai (2010).

Usage

CBUM(p, start.pi0=0.5, thresh.censor=0.05, eps=1e-5, niter=Inf, verbose=FALSE)

Arguments

p	a numeric vector the p-values
start.pi0	numeric scalar, starting value of pi0 for EM algorithm.
thresh.censor	numeric scalar, the threshold of censoring. If isTRUE(lambda<min(p)), this is equivalent to the BUM method of Pounds and MOrris (2003).
eps	numeric scalar, maximum tolerable absolute difference of parameter estimates for successive iterations in the EM algorithm.
niter	numeric scalar, maximum number of EM iterations.
verbose	logical scalar, indicating whether excessive outputs will be printed during EM algorithm.

Details

This function is an improved version of the CBpi0 function available at http://home.gwu.edu/~ylai/research/CBpi0/CBpi0.txt, which implements the censored (1-paramater beta)-uniform mixture model to a large number of p-values.

Value

A numeric scalar, being the proportion pi0 of true null hypotheses. The result has a class 'CBUM', with the following attributes:

converged	logical, convergence status.
iter	numeric, number of iterations.
call	the match.call() result.
alpha	estimated alpha parameter for the beta component.
lfdr	numeric vector estimated local false discovery rates, if thresh.censor<min(p); NULL, otherwise
thresh.censor	the censoring threshold for p-values.

Author(s)

Long Qu modified the code from Markitsis and Lai (2010).

References

Anastasios Markitsis and Yinglei Lai (2010) A censored beta mixture model for the estimation of the proportion of non-differentially expressed genes. Bioinformatics 26(5):640-646.

See Also

qvalue, histf1

Examples

set.seed(99722)
p=c(runif(3500), rbeta(1500, .8, 2.2))
CBUM(p)

gitlongor/pi0 documentation built on May 17, 2019, 5:29 a.m.