plrs.series: Fit plrs models for a series of arrays.

In plrs: Piecewise Linear Regression Splines (PLRS) for the association between DNA copy number and gene expression

Description

The function fits plrs models for a series of arrays. Model selection and testing procedures may be applied.

Usage

plrs.series(expr, cghseg, cghcall=NULL,
probloss = NULL, probnorm = NULL, probgain = NULL, probamp = NULL,
control.model  = list(continuous = FALSE,
                      constr = TRUE,
                      constr.slopes = 2,
                      constr.intercepts = TRUE,
                      min.obs = 3,
                      discard.obs = TRUE),
control.select = list(crit = ifelse(control.model$constr, "osaic","aic")),
control.test   = list(testing = TRUE,
                      cb = FALSE,
                      alpha = 0.05),
control.output = list(save.models = FALSE,
                      save.plots = FALSE,
                      plot.lin = FALSE,
                      type = "jpeg"))

Arguments

expr	Either a matrix of expression profiles or an ExpressionSet object.
cghseg	Either a matrix of segmented copy number values or objects of class cghSeg or cghCall
cghcall	Matrix of called copy number
probloss	Matrix of call probabilities associated with state "loss". Default is NULL.
probnorm	Matrix of call probabilities associated with state "normal". Default is NULL.
probgain	Matrix of call probabilities associated with state "gain". Default is NULL.
probamp	Matrix of call probabilities associated with state "amplification". Default is NULL.
control.model	See details
control.select	See details
control.test	See details
control.output	See details

Details

If DNA and mRNA input data are matrices, rows should correspond to genes and columns to arrays. Alternatively, expression data may be provided as an ExpressionSet object and aCGH data as cghSeg or cghCall objects. A cghCall object contain all data from the calling step, thus arguments probloss, probnorm, probnorm and probamp can be omitted. An object of class cghSeg does not contain such data so only simple linear models will be fitted.

control.model allows the user to specify the type of model that has to be fitted. This must be a list with one or more of the following components: constr, constr.slopes, constr.intercepts, min.obs and discard.obs. See functions plrs and modify.conf for more details.

control.select allows the user to specify whether model selection should be done and how. This must be a list with a component named crit. See function plrs.select for more details. If control.select = NULL then no model selection is done.

control.output allows the user to plot and save each plrs model. This must be a list with components:
save.models, a logical. This will create within the work directory a new directory named "plrsSeriesObjects" that will contain all objects.
save.plots, a logical. This will create within the work directory a new directory named "plrsSeriesPlots" that will contains all saved plots.
plot.lin, a logical. Whether the simple linear model should aslo be plotted.
type, a character. Format of file. To pass through function savePlot.

Value

An object of class plrs.series-class

Author(s)

Gwenael G.R. Leday [e-mail]

Examples

# Simulate data
ngenes <- 10
narray <- 48
rna <- dnaseg <- dnacal <- matrix(NA, ngenes, narray)
idx <- sample(1:4, ngenes, replace=TRUE, prob=rep(1/4,4))
for(i in 1:ngenes){
	Sim <- plrs.sim(n=narray, states=idx[i], sigma=0.5)
	rna[i,] <- Sim$expr
	dnaseg[i,] <- Sim$seg
	dnacal[i,] <- Sim$cal
}


# Screening procedure with linear model
series <- plrs.series(expr = rna, cghseg = dnaseg, cghcall = NULL, control.select = NULL)

# Screening procedure with full plrs model
series <- plrs.series(expr = rna, cghseg = dnaseg, cghcall = dnacal, control.select = NULL)

# Model selection
series <- plrs.series(expr = rna, cghseg = dnaseg, cghcall = dnacal)

plrs documentation built on April 28, 2020, 6:09 p.m.