pumaClust {puma}R Documentation

Propagate probe-level uncertainty in model-based clustering on gene expression data

Description

This function clusters gene expression using a Gaussian mixture model including probe-level measurement error. The inputs are gene expression levels and the probe-level standard deviation associated with expression measurement for each gene on each chip. The outputs is the clustering results.

Usage

pumaClust(e=NULL, se=NULL, efile=NULL, sefile=NULL, 
      subset=NULL, gsnorm=FALSE, clusters, 
      iter.max=100, nstart=10, eps=1.0e-6, del0=0.01)

Arguments

e either a valid ExpressionSet object, or a data frame containing the expression level for each gene on each chip.
se data frame containing the standard deviation of gene expression levels.
efile character, the name of the file which contains gene expression measurements.
sefile character, the name of the file which contains the standard deviation of gene expression measurements.
subset vector specifying the row number of genes which are clustered on.
gsnorm logical specifying whether do global scaling normalisation or not.
clusters integer, the number of clusters.
iter.max integer, the maximum number of iterations allowed in the parameter initialisation.
nstart integer, the number of random sets chosen in the parameter initialisation.
eps numeric, optimisation parameter.
del0 numeric, optimisation parameter.

Details

The input data is specified either as an ExpressionSet object (in which case se, efile and sefile will be ignored), or by e and se, or by efile and sefile.

Value

The result is a list with components
cluster: vector, containing the membership of clusters for each gene; centers: matrix, the center of each cluster; centersigs: matrix, the center variance of each cluster; likelipergene: matrix, the likelihood of belonging to each cluster for each gene; bic: numeric, the Bayesian Information Criterion score.

Author(s)

Xuejun Liu, Magnus Rattray

References

Liu,X., Lin,K.K., Andersen,B., and Rattray,M. (2006) Propagating probe-level uncertainty in model-based gene expression clustering, technical report available upon request.

Liu,X., Milo,M., Lawrence,N.D. and Rattray,M. (2005) A tractable probabilistic model for Affymetrix probe-level analysis across multiple chips, Bioinformatics, 21(18):3637-3644.

See Also

Related method mmgmos

Examples

  data(Clust.exampleE)
  data(Clust.exampleStd)
  pumaClust.example<-pumaClust(Clust.exampleE,Clust.exampleStd,clusters=7)

[Package puma version 1.8.1 Index]