p0est: Estimates the proportion of unaffected features (p0).

In PsiHat: Several Local False Discovery Rate Estimators

Description

Estimates the proportion of null hypothesis (unaffected features) (p0). It is a wrapper of function qvalue from R package qvalue.

Usage

p0est(pvalue, lambda = seq(0, 0.9, 0.05), pi0.method = "smoother", smooth.df = 3,
    smooth.log.pi0 = FALSE, fdr.level = NULL)

Arguments

pvalue

Vector of p-values.

lambda	Tuning parameter in [0,1) to estimate p0.
pi0.method	Method to estimate the proportion of true null hypothesis (p0): "smoother" or "bootstrap".
smooth.df	If pi0.method="smoother", number of degrees-of-freedom.
smooth.log.pi0	If pi0.method="smoother" and smooth.log.pi0=TRUE, p0 is estimated by applying a smoother to log p0 estimates against the tuning parameter lambda.
fdr.level	A level at which to control the FDR. Must be in (0,1]

.

Value

A number in [0,1]

Author(s)

Code: Marta Padilla (wrapper of function qvalue from R package qvalue)
Documentation: Alaa Ali, Kyle Leckett, Marta Padilla.

See Also

qvalue package, lfdr.bbe and nqvalue.

Examples

# a p-value with missing elements:
y <- runif(15) ;y[c(13, 2,5)] <- NA
z1 <- p0est(pvalue = y, lambda = 0.5)
z2 <- p0est(pvalue = y, pi0.method = "bootstrap")

PsiHat documentation built on May 30, 2017, 7:12 a.m.