estimate_IG_indicators {BUScorrect} | R Documentation |
Call the function to estimate the intrinsic gene indicators.
estimate_IG_indicators(BUSfits, postprob_DE_threshold = 0.5)
BUSfits |
The BUSfits object output by the function |
postprob_DE_threshold |
the threshold to call an intrinsic gene indicator to be one or not according to whether its posterior probability is higher than postprob_DE_threshold or not. The default is 0.5. |
est_L |
the estimated intrinsic gene indicators, a matrix where the rows represent genes and the columns correspond to subtypes k=2,...,K |
Xiangyu Luo
Xiangyu Luo, Yingying Wei. Batch Effects Correction with Unknown Subtypes. Journal of the American Statistical Association. Accepted.
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 the posterior probability threshold to estimate the 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)