m2loglik.normexp {limma}R Documentation

Normal + Exponential Log-Likelihood

Description

Marginal log-likelihood of foreground values for normal + exponential model and its derivatives. This function is called by backgroundCorrect and is not normally called directly by the user.

Usage

m2loglik.normexp(theta,foreground,background=NULL,background.matrix=NULL)
sumloglik(theta,foreground,background)
grsumloglik(theta,foreground,background)

Arguments

theta numeric vector of parameters
foreground numeric vector of foreground intensities
background optional vector of background intensity values
background.matrix option design matrix for regression on background values

Details

Computes minus twice the log-likelihood based on the $normal(μ,σ^2)+exponential(α)$ convolution model for the foreground intensities. The last two elements of theta are $log(σ)$ and $log(α)$. The other elements are the vector $β$.

The vector $β$ is related to $μ$ through $μ=Xβ$ if $X=$background.matrix is non-null. Otherwise $μ=β_1+β_2$background if background is non-null. If both background and background.matrix are null, then $μ=β_1$.

m2loglik.normexp is minus-twice the log-likelihood. sumloglik is minus the log-likelihood. grsumloglik is the gradient vector and grloglik the gradient components for individual spots.

Value

Numeric scalar giving minus twice the log-likelihood

Bugs

There are undocumented arguments in the function grloglik. These will disappear soon.

Author(s)

Jeremy Silver and Gordon Smyth

See Also

fit.normexp, signal.normexp

An overview of normalization and background correction functions is given in 4.Normalization.


[Package limma version 1.9.6 Index]