Merge pull request #137 from gdrplatform/GDR-2736

Gdr 2736

DESCRIPTION

@@ -1,8 +1,8 @@
 Package: gDRutils
 Type: Package
 Title: A package with helper functions for processing drug response data
-Version: 1.3.16
-Date: 2024-10-11
+Version: 1.3.17
+Date: 2024-10-24
 Authors@R: c(person("Bartosz", "Czech", role=c("aut"),
                    comment = c(ORCID = "0000-0002-9908-3007")),
              person("Arkadiusz", "Gladki", role=c("cre", "aut"), email="[email protected]",

NEWS.md

@@ -1,3 +1,6 @@
+## gDRutils 1.3.17 - 2024-10-24
+* make average_biological_replicates_dt more restrictive - fewer columns to check for duplicates
+
 ## gDRutils 1.3.16 - 2024-10-11
 * make duplicates' helpers supporting combo assays as well
 

R/headers_list.R

@@ -204,6 +204,17 @@
       "Fit Type GR",
       "RV_fit_type",
       "GR_fit_type"
+   ),
+   # due to the fact that there is some freedom in what values are in individual fields, 
+   # in order to avoid duplicates in the application we have to exclude some fields from 
+   # recognizing duplicates in averaging
+   blacklisted = c(
+     "Tissue", # sometimes this field is missing
+     "cellline_tissue",
+     "Reference Division Time", # sometimes this field has `NA`s
+     "cellline_ref_div_time",
+     "Parental Identifier", # sometimes suffixes incorrectly differentiate this field
+     "cellline_parental_identifier"
    )
   )
 }

R/utils.R

@@ -435,6 +435,8 @@ geometric_mean <- function(x, fixed = TRUE, maxlog10Concentration = 1) {
 #' to take the geometric average of.
 #' @param fit_type_average_fields Character vector of column names in \code{dt} 
 #' that should be treated as a column with fit type data
+#' @param blacklisted_fields Character vector of column names in \code{dt} 
+#' that should be skipped in averaging
 #' @param add_sd Flag indicating whether to add standard deviation and count columns.
 #' 
 #' @examples
@@ -452,6 +454,7 @@ average_biological_replicates_dt <- function(
     fixed = TRUE,
     geometric_average_fields = get_header("metric_average_fields")$geometric_mean,
     fit_type_average_fields = get_header("metric_average_fields")$fit_type,
+    blacklisted_fields = get_header("metric_average_fields")$blacklisted,
     add_sd = FALSE) {
 
   checkmate::assert_data_table(dt)
@@ -476,7 +479,8 @@ average_biological_replicates_dt <- function(
   average_fields <- setdiff(names(Filter(is.numeric, data)), c(unlist(pidfs), var, iso_cols))
   geometric_average_fields <- intersect(geometric_average_fields, names(dt))
   fit_type_average_fields <- intersect(fit_type_average_fields, names(dt))
-  group_by <- setdiff(names(data), c(average_fields, var, id_cols, fit_type_average_fields))
+  blacklisted_fields <- intersect(blacklisted_fields, names(dt))
+  group_by <- setdiff(names(data), c(average_fields, var, id_cols, fit_type_average_fields, blacklisted_fields))
 
   if (add_sd) {
     # Calculate standard deviation for both average_fields and geometric_average_fields

tests/testthat/test-utils.R

@@ -252,7 +252,6 @@ test_that("average_biological_replicates_dt works as expected", {
   av1i <- average_biological_replicates_dt(tdata, var = "source_id", fit_type_average_fields = "bad_value")
   expect_true(nrow(av1i) == 8)
 
-
 })
 
 test_that("has_single_codrug_data works as expected", {