R: Estimate hyperparameters for DE model using expression...

estimateHyperPar {BitSeq}

R Documentation

Estimate hyperparameters for DE model using expression samples and joint mean expression

Description

Estimate hyperparameters for the Differential Expression model using expression samples and produced smoothed values of the hyperparameters depending on joint mean expression.

Usage

estimateHyperPar( outFile, conditions=NULL, paramsInFile=NULL, 
      meanFile=NULL, force=TRUE, exThreshold=NULL, lambda0=NULL, 
      paramsAllFile=NULL, smoothOnly=NULL, lowess_f=NULL, lowess_steps=NULL,
      verbose=NULL, veryVerbose=NULL, norm=NULL, seed=NULL, pretend=FALSE )

Arguments

`outFile`	Name of the output file.
`conditions`	`List` of vectors, each vector containing names of files containing the expression samples from a replicate (Can be both technical and biological replicates. However, in order to get good results biological replicates for each condition are essential).
`paramsInFile`	File produced by previous run of the function using `paramsAllFile` flag.
`meanFile`	Name of the file containing joint mean and variance.
`exThreshold`	Threshold of lowest expression for which the estimation is done.
`paramsAllFile`	Name of the file to which to store all parameter values generated prior to lowess smoothing(good for later, more careful re-smoothing.)
`smoothOnly`	Input file contains previously sampled hyperparameters which should smoothed only.
`verbose`	Verbose output.

Advanced options:

`force`	Force smoothing hyperparameters, otherwise program might not produce parameters file at the end.
`lambda0`	Model parameter lambda0.
`lowess_f`	Parameter F for lowess smoothing specifying amount of smoothing.
`lowess_steps`	Parameter Nsteps for lowess smoothing specifying number of iterations.
`veryVerbose`	More verbose output.
`norm`	Vector of (multiplicative) normalization constants for library size normalization of expression samples. Number of constants has to match the number of expression samples files.
`seed`	Sets the initial random seed for repeatable experiments.
`pretend`	Do not execute, only print out command line calls for the C++ version of the program.

Value

`.par`	file containing the smoothed hyperparameters
`.ALLpar`	file containing all hyperparameter samples prior to smoothing - optional

Author(s)

Peter Glaus

Examples

## Not run: 
cond1Files = c("data-c0b0.rpkm","data-c0b1.rpkm")
cond2Files = c("data-c1b0.rpkm","data-c1b1.rpkm")
estimateHyperPar( conditions=list(cond1Files, cond2Files), outFile="data.par", 
      meanFile="data.means", verbose=TRUE)

estimateHyperPar( conditions=list(cond1Files, cond2Files), outFile="data.par", 
      meanFile="data.means", paramsFile="data.ALLpar", force=FALSE)
estimateHyperPar( outFile="data.par", paramsInFile="data.ALLpar", smoothOnly=TRUE )

## End(Not run)

[Package BitSeq version 1.26.1 Index]