fitBetaUniformMixtureDistribution: Fit a Beta-Uniform decomposition model to the provided...

In adamsardar/metaDEGth: Combining differential expression P-values in a statistically consistent fashion whilst respecting background signal/noise trends

Fit a Beta-Uniform decomposition model to the provided p-values

Description

Following the approach of Morris & Pounds, a distribution of P-values can be decomposed as a mixture of two beta distributions, one with shape paramter (α) of 1 - modelling noise under the null-hypothesis - and another with a variable (a), modelling signal. One can view this as modeling P-values as a random process where P ~ (1-λ)β(a,1) + λβ(1,1)

Usage

fitBetaUniformMixtureDistribution(pValues, nStarts = 10L)

## S4 method for signature 'Pvalues,ScalarInteger'
fitBetaUniformMixtureDistribution(pValues, nStarts = 10L)

## S4 method for signature 'numeric,ANY'
fitBetaUniformMixtureDistribution(pValues, nStarts = 10L)

## S4 method for signature 'Pvalues,numeric'
fitBetaUniformMixtureDistribution(pValues, nStarts = 10L)

Arguments

pValues	A numeric vector of P-values for which to perform Beta-Uniform decomposition
nStarts	How many repeats of the fitting routine should the routine run before the best LLH is picked? The default is 10, but in some cases (when there is no 'clean' P-value distribution), it might be worth running for more iterations.

Details

Analagous to BioNet::fitBumModel.

Value

A BetaUniformModel object, detailign the fit

Functions

• fitBetaUniformMixtureDistribution,Pvalues,ScalarInteger-method: Typed method for dispatch

• fitBetaUniformMixtureDistribution,numeric,ANY-method: Adapter method, attempts to convert numeric values to Pvalues

• fitBetaUniformMixtureDistribution,Pvalues,numeric-method: fitting using nStarts defined by the user as numeric

References

Pounds, S., & Morris, S. W. (2003). Estimating the occurrence of false positives and false negatives in microarray studies by approximating and partitioning the empirical distribution of p-values. Bioinformatics

See Also

betaUniformScore BioNet::fitBumModel

Examples

library(ggplot2)
data(siTAZdiffex_HFL1_diffexDT)

siTAZ1_BetaUniformMod <- fitBetaUniformMixtureDistribution(siTAZdiffex_HFL1_diffexDT$siTAZ1_pValue)

siTAZ1_BetaUniformMod

plot(siTAZ1_BetaUniformMod) # Produce some diagnostic plots

betaUnifScored <- betaUniformScore(siTAZ1_BetaUniformMod, FDR = 0.05)
qplot(betaUnifScored)

adamsardar/metaDEGth documentation built on Sept. 28, 2022, 10:12 p.m.