postprob_DE_thr_fun: Select the the posterior probability threshold to control the...

In BUScorrect: Batch Effects Correction with Unknown Subtypes

Description

To control the false discovery rate at the targeted level, call postprob_DE_thr_fun to obtain the threshold for the posterior probability of being differentially expressed.

Usage

postprob_DE_thr_fun(BUSfits, fdr_threshold = 0.1)

Arguments

BUSfits	The BUSfits object output by the function BUSgibbs.
fdr_threshold	the false discovery rate level we want to control.

Value

thre0

the posterior probability threshold that controls the false discovery rate.

Author(s)

Xiangyu Luo

References

Xiangyu Luo, Yingying Wei. Batch Effects Correction with Unknown Subtypes. Journal of the American Statistical Association. Accepted.

Examples

rm(list = ls(all = TRUE))  
set.seed(123)
#a toy example, there are 6 samples and 20 genes in each batch
example_Data <- list()

#batch 1
example_Data[[1]] <- rbind(matrix(c(1,1,5,5,10,10,
						3,3,7,7,12,12), ncol=6, byrow=TRUE), matrix(c(1,2),nrow=18, ncol=6))

#batch 2
batch2_effect <- c(2,2,2,1,1)
example_Data[[2]] <- rbind(matrix(c(1,1,5,5,10,10,
						3,3,7,7,12,12), ncol=6, byrow=TRUE), matrix(c(1,2),nrow=18, ncol=6)) + batch2_effect

#batch 3
batch3_effect <- c(3,2,1,1,2)
example_Data[[3]] <- rbind(matrix(c(1,1,5,5,10,10,
						3,3,7,7,12,12), ncol=6, byrow=TRUE), matrix(c(1,2),nrow=18, ncol=6)) + batch3_effect
set.seed(123)
BUSfits <- BUSgibbs(example_Data, n.subtypes = 3, n.iterations = 100, showIteration = FALSE)
#select kappa to estimate intrinsic gene indicators
thr0 <- postprob_DE_thr_fun(BUSfits, fdr_threshold=0.1)
est_L <- estimate_IG_indicators(BUSfits, postprob_DE_threshold=thr0)

#obtain the intrinsic gene indicators
intrinsic_gene_indices <- IG_index(est_L)

BUScorrect documentation built on Nov. 8, 2020, 8:06 p.m.