pack {vabayelMix}R Documentation

Profile analysis using Clustering and Kurtosis

Description

For a data matrix, selects features with kurtosis values in a specified range. Optionally, it further selects features based on whether their profiles are mixtures of gaussians or not.

Usage

 pack(data.m,kurt.range=c(-2,0),cluster=T,method=c("bic","vb"));

Arguments

data.m Data matrix with features along the rows.
kurt.range Desired range of kurtosis values.
cluster Logical, to indicate whether additional cluster learning step is desired.
method Character specifying model selection to be used (bic=EM-algorithm + BIC, vb=variational Bayesian + evidence bound).

Value

A list with the following components:

out Matrix with rows labeling selected features and columns labeling kurtosis, cluster size and index position in data.m
class A list with non-null elements giving the clustering classification of the selected features.

Author(s)

Andrew Teschendorff aet21@hutchison-mrc.cam.ac.uk


[Package vabayelMix version 0.3 Index]