call.peaks: Call peaks from a list of window-level p-values

In BinQuasi: Analyzing Replicated ChIP Sequencing Data Using Quasi-Likelihood

Description

Call peaks from a list of p-values corresponding to window-level tests on a genomic partition of ChIP-seq counts. Used within the main peak calling function, BQ.

Usage

call.peaks(window.pvals, method = c("BY", "BH", "none"), start, end,
  chromosomes, alpha = 0.05)

Arguments

window.pvals	Vector of p-values, with each element corresponding to a window of a genomic partition. Typically obtained from the QL.fit and QL.results functions.
method	Correction method applied to window.pvals. Must be one of "BH", "BY", or "none" to specify Benjamini-Hochberg, Benjamini-Yekutieli, or no adjustment, respectively.
start	Vector of the genomic start locations corresponding to the supplied p-values.
end	Vector of the genomic end locations corresponding to the supplied p-values.
chromosomes	Vector of the chromosome names corresponding to the supplied p-values.
alpha	The desired significance threshold in (0, 0.5).

Details

After correcting for multiple testing using the adjustment specified by method, windows that are significant according to the threshold alpha are merged if adjacent and retained as candidate regions. Simes' procedure is used to combine the window-level p-values in each candidate region into a region-level p-value. The Benjamini-Hochberg procedure is applied to the resulting candidate regions and those that exceed the significance threshold alpha are returned as peaks.

Value

The called peaks as a dataframe with variables:

start	Genomic start locations of the called peaks.
end	Genomic end locations of the called peaks.
width	Width of the called peaks.
chr	Chromosomes of the called peaks.
P.val	p-values of the called peaks (aggregated from the windows comprising the peak using Simes' procedure).
Q.val	q-values of the called peaks (computing using the Benjamini-Hochberg procedure).

Author(s)

Emily Goren ([e-mail]).

References

Benjamini and Hochberg (1995) "Controlling the false discovery rate: a practical and powerful approach to multiple testing" Journal of the Royal Statistical Society Series B, 57: 289-300.

Benjamini and Yekutieli (2001) "The control of the false discovery rate in multiple testing under dependency" Annals of Statistics. 29: 1165-1188.

Simes (1986) "An improved Bonferroni procedure for multiple tests of significance" Biometrika, 73(3): 751-754.

Examples

# Example for a single chromosome.
start <- seq(1, 1e6, by = 200)
end <- start + 200 - 1
chromosomes <- rep('chr1', length(start))
p <- c(runif(length(start) - 10), rep(1e-12, 10))
called <- call.peaks(p, "BH", start, end, chromosomes)
called

BinQuasi documentation built on May 2, 2019, 4:16 a.m.