hybridization {MixThres}R Documentation

Hybridization : estimation and selection of the mixture model

Description

This function estimates the probe intensity values distribution using a particular mixture model in order to define in a second step a hybridization threshold.

Usage

hybridization(x, leftT = c(TRUE, FALSE, TRUE, FALSE), rightT = c(TRUE, TRUE, FALSE, FALSE), stop.crit = 1e-06, Kmax = 5)

Arguments

x A dataframe containing a vector of probenames and a vector of probe intensity values.
leftT A vector of boolean specifying the presence of a left truncature.
rightT A vector of boolean specifying the presence of a right truncature of the same length than leftT.
stop.crit The stopping relative precision limit for stopping EM algorithm.
Kmax The maximal number of components in the mixture model.

Details

Truncated gaussian distributions are used. An EM algorithm with a supplement M-step corresponding to a fixed-point algorithm (in order to correct bias) is used to estimate the parameters of the mixture. A vector epsilon algorithm is used to accelerate the EM algorithm. The model is selected according to the BIC.

Value

An object of class MixThres,

GeneID The vector of probenames.
InputData The vector of input intensity values.
ModelInfo A list concerning the selected model : nb of components, parameters estimation, loglikelihood, nb of parameters, aic, bic
Prob The matrix of posterior probabilities.
Trunc A vector of the two truncatures values.
Bic A vector of the BIC calculated for all the models.
NbParam A vector of the number of parameters for all the models.
Crit NULL. Will be filled with the use of the threshold function.
Threshold NULL. Will be filled with the use of the threshold function.
Hybrid NULL. Will be filled with the use of the threshold function.

Author(s)

J. Aubert

References

Picard et al., Using mixture models to define a hybridization threshold in DNA microarray experiments.

See Also

threshold, print.MixThres, plot.MixThres, hist.MixThres

Examples

## Data simulation
Signal <- c(rnorm(100,5,1),rnorm(700,8,1),rnorm(200,12,1))
fileIN <- data.frame(Gene=paste("Gene",1:1000,sep=""),Signal=Signal)

## Mixture models
out <- hybridization(fileIN)

[Package MixThres version 1.0 Index]