Minor changes

DESCRIPTION

@@ -1,10 +1,10 @@
 Package: PAMES
-Date: 2021-03-12
+Date: 2021-12-01
 Type: Package
 Title: Purity Assessment from clonal MEthylation Sites
 Description: Exploiting data from DNA methylation, this package provides the operations
     required to evaluate the purity of tumor samples.
-Version: 2.7.1
+Version: 2.7.2
 Authors@R: c(person("Dario", "Romagnoli", role=c("aut", "cre"),
   email="[email protected]"),
   person("Matteo", "Benelli", role="aut"),
@@ -24,5 +24,5 @@ Suggests:
 License: GPL-3 | file LICENSE
 Encoding: UTF-8
 LazyData: true
-RoxygenNote: 7.1.1
+RoxygenNote: 7.2.0
 VignetteBuilder: knitr

NEWS.md

@@ -1,3 +1,6 @@
+# PAMES v2.7.2
+- rename `median_of_regions`: `reduce_region`
+- add `method` parameter to `reduce_to_regions`: allow choice between "median" (default) or mean
 # PAMES v2.7.1
 - fix `select_informative_regions_ext` with flag `return_info=TRUE`
 # PAMES v2.7.0

R/reduce_to_regions.R

@@ -11,14 +11,17 @@
 #' @param min_CpGs An integer (default to 3). Minimum number of CpG sites
 #' within a single genomic region required to compute the reduced beta value
 #' (return NA otherwise).
+#' @param method Take `median` or `mean`of CpG sites.
 #' @return A matrix of beta values (nrow == length(cpg_indexes)).
 #' @importFrom stats median
 #' @export
 #' @examples
 #' reduced_data <- reduce_to_regions(bs_toy_matrix, bs_toy_sites, cpg_islands[1:10,])
-reduce_to_regions <- function(beta_table, cpg_sites, cpg_regions, min_CpGs = 3){
+reduce_to_regions <- function(beta_table, cpg_sites, cpg_regions, min_CpGs = 3, method=c("median", "mean")){
+
     message(sprintf("[%s] # Reduce to regions #", Sys.time()))
     # check parameters
+    method <- match.arg(method)
     min_CpGs <- as.integer(min_CpGs)
     assertthat::assert_that(min_CpGs > 0)
     assertthat::assert_that(ncol(cpg_sites) >= 2)
@@ -63,7 +66,7 @@ reduce_to_regions <- function(beta_table, cpg_sites, cpg_regions, min_CpGs = 3){
     reduced_data <- lapply(seq_along(idx_list[above_thr_regions]), function(i) {
             utils::setTxtProgressBar(pb, i)
             idx <- idx_list[[i]]
-            median_of_region(beta_table[idx,,drop = FALSE], min_CpGs)
+            summarise_region(beta_table[idx,,drop = FALSE], min_CpGs, method)
     })
     close(pb)
 
@@ -73,20 +76,23 @@ reduce_to_regions <- function(beta_table, cpg_sites, cpg_regions, min_CpGs = 3){
     return(reduced_table)
 }
 
-#' Transform CpG sites to one CpG region
+#' Reduce many CpG sites to one CpG region
 #'
-#' If the number of sites is sufficient take the median value else return NA.
+#' If the number of sites is sufficient, take the median/mean value else return NA.
 #' @param x A subset matrix.
 #' @param n Minimum required number of sites per region (return NA otherwise).
+#' @param method Either `median` or `mean`.
 #' @return A vector
 #' @keywords internal
-median_of_region <- function(x, n) {
-    # remove sites non reported for all samples
+summarise_region <- function(x, n, method) {
+    # remove fully NA sites
     valid_sites <- which(rowSums(is.na(x)) != ncol(x))
     x <- x[valid_sites,,drop=FALSE]
     if (nrow(x) < n) {
         return(rep(NA, ncol(x)))
-    } else {
+    } else if (method=="median"){
         return(apply(x, 2, median, na.rm = TRUE))
+    } else if (method=="mean"){
+        return(apply(x, 2, mean, na.rm = TRUE))
     }
 }

tests/testthat/test-PAMES.R

@@ -33,8 +33,8 @@ context("CpG regions") ##################################################
 test_that("median of regions", {
   x <- rbind(rep(NA, 10), sample(100, 10), sample(100, 10))
   x[,3] <- NA
-  expect_type(median_of_region(x, 2), "double")
-  expect_true(all(is.na(median_of_region(x, 3))))
+  expect_type(summarise_region(x, 2, "median"), "double")
+  expect_true(all(is.na(summarise_region(x, 3, "mean"))))
 })
 test_that("reduce_to_regions works", {
   expect_error(reduce_to_regions(tumor_toy_data, illumina27k_hg38[3:4], cpg_islands), "No shared chromosomes")