Return a profile likelihood function for further use.

A profile likelihood function returned by the pl considers parameters of other genes. In contrast, the one from the plGene assumes parameters of other genes, shared parameters and the normalisation factors to be fixed.
The profile likelihood is estimated by numerical optimisation, and it can be sensitive to the initial values. The optimisation is repeated options$replicates times (default: 10), by adding a random values to the the optimum parameters at a scale specified in the options$jitter.

plGene(parName, geneIndex, par, pd, options)

pl(paramPath, par, pd, options, freeParams = names(par))

Arguments

parName	a character, e.g. "mu"
geneIndex	an integer(1); a row index which corresponds to the investigating gene
par	a result of the fitModel function
pd	a PulseData object
options	an option list used for the fitModel call; additional options can be specified: jitter, a double (default: .1) replicates, an integer number of repeating the optimisation from random points absolute, a logical (default: FALSE); if FALSE, the likelihood value at the optimal point (i.e. par) is substracted from the returned value. In this case, the value of the returned function at the point par is 0.
paramPath	a list with names and indexes in order to locate the parameter in the par argument (e.g. list("mu", 1) corresponds to the "mu" parameter value for the first gene, i.e. par[["mu"]][[1]].
freeParams	which parameters are optimised (i.e. not fixed); by default they are derived from the names of the boundaries

Value

a function with the calling convention as f(x = double(1)) --> double(1)), which return the value of -log(likelihood(mu = x))

Details

The randomisation of the parameter x is made as $ x^* = (1 + a) * x$, where $a$ is a random value from the uniform distribution (0,options$jitter).