biomarkerCategorization: biomarker categrorization

In AWFisher: An R package for fast computing for adaptively weighted fisher's method

Description

biomarker categrorization

Usage

biomarkerCategorization(studies, afunction, B = 10, DEindex = NULL,
  fdr = NULL, silence = FALSE)

Arguments

studies	a list of K studies. Each element (kth study) of the list is another list consisting gene expression matrix and and label information.
afunction	A function for DE analysis. Options can be function_limma or function_edgeR. Default option is function_limma. However, use could define their own function. The input of afunction should be list(data, label) which is consistent with one element of the studies list/argument. The return of afunction should be list(pvalue=apvalue, effectSize=aeffectsize)
B	number of permutation should be used. B=1000 is suggested.
DEindex	If NULL, BH method will be applied to p-values and FDR 0.05 will be used. User could specify a logical vector as DEindex.
fdr	Default is 0.05. The co-membership matrix calculation will base on genes with this specified fdr.
silence	If TRUE, will print out the bootstrapping procedure.

Details

biomarker categrorization via boostrap AW weight.

Value

A list consisting of biomarker categrorization result.

varibility	Varibility index for all genes
dissimilarity	Dissimilarity matrix of genes of DEindex==TRUE
DEindex	DEindex for Dissimilarity

Author(s)

Zhiguang Huo

Examples

N0 = 10
G <- 1000
GDEp <- 50
GDEn <- 50
K = 4

studies <- NULL
set.seed(15213)
for(k in seq_len(K)){
    astudy <- matrix(rnorm(N0*2*G),nrow=G,ncol=N0*2)
    ControlLabel <- seq_len(N0)
    caseLabel <- (N0 + 1):(2*N0)

    astudy[1:GDEp,caseLabel] <- astudy[1:GDEp,caseLabel] + 2
    astudy[1:GDEp + GDEn,caseLabel] <- astudy[1:GDEp + GDEn,caseLabel] - 2

    alabel = c(rep(0,length(ControlLabel)),rep(1,length(caseLabel)))

    studies[[k]] <- list(data=astudy, label=alabel)
}


result <- biomarkerCategorization(studies,function_limma,B=100,DEindex=NULL)
sum(result$DEindex)
head(result$varibility)
print(result$dissimilarity[1:4,1:4])

AWFisher documentation built on Nov. 8, 2020, 5:42 p.m.