ChIPseqR (version 1.26.0)

getCutoff: Determine significance threshold for binding site scores

Description

Given a vector of observed binding site scores and a desired false discovery rate, this function returns the lowest score that should be considered significant to achieve the given false discovery rate.

Usage

getCutoff(score, alpha = 0.05, tailCut = 0.95, adapt = FALSE, lambda, plot = TRUE, returnPval = TRUE)

Arguments

score

Numeric vector with binding site scores.

alpha

Desired false discovery rate.

tailCut

Truncation point used to exclude outliers when fitting the null distribution.

adapt

Logical indicating whether an adaptive false discovery rate should be used.

lambda

If adapt is TRUE this is used in estimating the proportion of scores that is unrelated to binding sites.

plot

If this is TRUE (the default) a plot of the observed score distribution and estimated null distribution is generated.

returnPval

Indicates whether or not the corrected p-values for all scores should be returned.

Value

A list with components

cutoff

A numeric vector with the score and nominal false discovery rate corresponding to the determined cutoff.

h0

A numeric vector with the mean and standard deviation of the estimated null distribution.

pvalue

If returnPval is TRUE, the p-values corresponding to the scores in score. Note that all missing values are removed.

pi0

If adapt is TRUE, the estimated proportion of scores not related to binding sites.

References

For the adaptive false discovery rate procedure used if adapt=TRUE see JD Storey, JE Taylor and D Siegmund. Strong control, conservative point estimation and simultaneous conservative consistency of false discovery rates: a unified approach. Journal of the Royal Statistical Society B, 66(1):187-205, 2004.

See Also

callBindingSites