DASC: Detecting hidden batch factors through data adaptive...

In DASC: Detecting hidden batch factors through data adaptive adjustment for biological effects

Description

Detecting hidden batch factors through data adaptive shrinkage and clustering (DASC)

Batch factor detection via DASC (Data-adaptive Shrinkage and Clustering-DASC)

Usage

DASC(edata, pdata, factor, method = "ama", type = 3, lambda, rank, nrun,
  spanning = FALSE, annotation)

Arguments

edata	the normalized target matrix, a data.frame The row is gene, the column is sample
pdata	Phenotypic data summarizes information about the samples
factor	A factor vector which controls the convex clustering
method	Algorithm to use: 'admm' or 'ama'
type	An integer indicating the norm used: 1 = 1-norm 2 = 2-norm 3 = 2-norm^2
lambda	A double number A regularization parameter in the convex optimization
rank	integer sequence
nrun	the iteration numbers of Semi-NMF
spanning	parameter is assigned as false
annotation	An annotation of the dataset

Details

The DASC function is the main function of our algorithm DASC (Data-adaptive Shrinkage and Clustering-DASC) package. The DASC includes two main steps

• Data-adaptive shrinkage using convex clustering shrinkage (Implemented by convex optimization.);

• Extract batch factors using matrix factorization.

Value

outputs the result of semi-NMF. It classifies each sample to its batch factor.

Author(s)

Haidong Yi, Ayush T. Raman

Haidong Yi, Ayush T. Raman

See Also

cvxclust_path_ama and cvxclust_path_admm for the detailed algorithm

Examples

library(NMF)
library(cvxclustr)
library(Biobase)
dat <- data.frame(matrix(rnbinom(n=200, mu=100, size=1/0.5), ncol=4))
pdat <- data.frame(sample = colnames(dat), type = c(rep('A',2), rep('B',2)))
rownames(pdat) <- colnames(dat)
res <- DASC(edata=dat, pdata=pdat, factor=pdat$type, method='ama', type=3,
lambda = 1, rank = 2, nrun = 50, spanning = FALSE,
annotation='simulated dataset')

DASC documentation built on Nov. 17, 2017, 9:36 a.m.