InfiniumClust {InfiniumPurify}R Documentation

Tumor sample clustering from Infinium 450k array data

Description

Clustering of tumor samples into subtypes accounting for tumor purity.

Usage

InfiniumClust(tumor.data, purity, K, maxiter = 100, tol = 0.001)

Arguments

tumor.data

numeric matrix of beta values for tumor samlpes. The rownames of tumor.data should be probe names of Infinium 450k array, and colnames should be names of tumor samples.

purity

purities for tumor samples. Could be estimated by getPurity, or user specified purities from other tools.

K

the number of clusters.

maxiter

the maximum number of iterations allowed. Default is 100.

tol

tolerance for convergence of EM iterations. Default is 0.001.

Details

An EM based statistical method for subtype classification based on DNA methylation data, while adjusting for tumor purity.

Value

InfiniumClust returns a list consisting oflikelihood tol.ll and membership matrix Z.

tol.ll

total log-likelihood of converged EM algorithm.

Z

the membership matrix, where row corresponds to tumor samples and column corresponds to K clusters.

Author(s)

Xiaoqi Zheng xqzheng@shnu.edu.cn and Hao Wu hao.wu@emory.edu

References

W. Zhang, H. Feng, H. Wu and X. Zheng (2016). Tumor purity improves cancer subtype classification from DNA methylation data. Submitted.

Examples

## load example data
data(beta.emp)
normal.data <- beta.emp[,1:21]
tumor.data <- beta.emp[,22:31]

## estimate tumor purity
purity <- getPurity(tumor.data = tumor.data,tumor.type= "LUAD")

## cluster tumor samples accounting for tumor purity
out <- InfiniumClust(tumor.data,purity,K=3, maxiter=5, tol=0.001)

[Package InfiniumPurify version 1.3.1 Index]