Merge pull request #135 from gdrplatform/GDR-2591

feat: make duplicates' helpers supporting combo assays as well

DESCRIPTION

@@ -1,8 +1,8 @@
 Package: gDRutils
 Type: Package
 Title: A package with helper functions for processing drug response data
-Version: 1.3.15
-Date: 2024-10-07
+Version: 1.3.16
+Date: 2024-10-11
 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 1.3.16 - 2024-10-11
+* make duplicates' helpers supporting combo assays as well
+
 ## gDRutils 1.3.15 - 2024-10-07
 * refactor the logic for dealing with duplicates in assay data
 

R/identifiers_list.R

@@ -43,7 +43,10 @@ IDENTIFIERS_LIST <- list(
   data_source = "data_source",
 
   # replicate
-  replicate = "Replicate"
+  replicate = "Replicate",
+
+  # normalization type
+  normalization_type = "normalization_type"
 )
 
 REQ_COL_IDENTIFIERS <- c(
@@ -79,7 +82,8 @@ EXPECT_ONE_IDENTIFIERS <- c(
   "drug_name3",
   "drug_moa3",
   "concentration3",
-  "data_source"
+  "data_source",
+  "normalization_type"
 )
 
 SYNONYMS_LIST <- list(

inst/settings.json

@@ -29,6 +29,7 @@
         "cellline_name",
         "concentration2",
         "duration",
-        "data_source"
+        "data_source",
+        "normalization_type"
     ]
 }

tests/testthat/test-duplicates.R

@@ -52,37 +52,51 @@ test_that("get_duplicated_rows works as expected", {
 })
 
 test_that("[has|get]_assay_dt_duplicated_rows works as expected", {
-
+
+  # single-agent data
   sdata <- get_synthetic_data("finalMAE_small")
-  smetrics_data <- convert_se_assay_to_dt(sdata[[1]], "Metrics")
-  smetrics_data_f <- gDRutils::flatten(
-    smetrics_data,
-    groups = c("normalization_type", "fit_source"),
-    wide_cols = gDRutils::get_header("response_metrics")
+  smetrics_data <- convert_se_assay_to_dt(sdata[[get_supported_experiments("sa")]], "Metrics")
+  smetrics_dup_data <- rbind(smetrics_data, smetrics_data[1:10, ])
+  expect_false(has_assay_dt_duplicated_rows(smetrics_data))
+  expect_true(has_assay_dt_duplicated_rows(smetrics_dup_data))
+
+  expect_equal(get_assay_dt_duplicated_rows(smetrics_dup_data),
+               c(1:10, 201:210))
+  expect_equal(
+    rbind(smetrics_data[1:10, ], smetrics_data[1:10, ]),
+    get_assay_dt_duplicated_rows(smetrics_dup_data, output = "data")
   )
-  expect_true(has_assay_dt_duplicated_rows(smetrics_data))
-  expect_false(has_assay_dt_duplicated_rows(smetrics_data_f))
+  expect_equal(get_assay_dt_duplicated_rows(smetrics_data), integer(0))
+  empty_dt <- get_assay_dt_duplicated_rows(smetrics_data, output = "data")
+  expect_true(nrow(empty_dt) == 0)
+  expect_is(empty_dt, "data.table")
 
-  expect_equal(get_assay_dt_duplicated_rows(smetrics_data), 1:200)
-  expect_equal(dim(smetrics_data), dim(get_assay_dt_duplicated_rows(smetrics_data, output = "data")))
-  expect_equal(get_assay_dt_duplicated_rows(smetrics_data_f), integer(0))
-  empty_dt <- get_assay_dt_duplicated_rows(smetrics_data_f, output = "data")
+  # combo data
+  cdata <- get_synthetic_data("finalMAE_combo_matrix_small")
+  cscores_data <- convert_se_assay_to_dt(cdata[[get_supported_experiments("combo")]], "scores")
+  cscores_dup_data <- rbind(cscores_data, cscores_data[1:10, ])
+  expect_false(has_assay_dt_duplicated_rows(cscores_data))
+  expect_true(has_assay_dt_duplicated_rows(cscores_dup_data))
+
+  expect_equal(get_assay_dt_duplicated_rows(cscores_dup_data),
+               c(1:10, 25:34))
+  expect_equal(
+    rbind(cscores_data[1:10, ], cscores_data[1:10, ]),
+    get_assay_dt_duplicated_rows(cscores_dup_data, output = "data")
+  )
+  expect_equal(get_assay_dt_duplicated_rows(cscores_data), integer(0))
+  empty_dt <- get_assay_dt_duplicated_rows(cscores_data, output = "data")
   expect_true(nrow(empty_dt) == 0)
   expect_is(empty_dt, "data.table")
 })
 
 test_that("throw_msg_if_duplicates works as expected", {
 
   sdata <- get_synthetic_data("finalMAE_small")
-  smetrics_data <- convert_se_assay_to_dt(sdata[[1]], "Metrics")
-  smetrics_data_f <- gDRutils::flatten(
-    smetrics_data,
-    groups = c("normalization_type", "fit_source"),
-    wide_cols = gDRutils::get_header("response_metrics")
-  )
-
+  smetrics_data <- convert_se_assay_to_dt(sdata[[get_supported_experiments("sa")]], "Metrics")
+  smetrics_dup_data <- rbind(smetrics_data, smetrics_data[1:10, ])
 
   exp_msg <- "rows are duplicated"
-  expect_error(throw_msg_if_duplicates(smetrics_data, "Metrics"), exp_msg)
-  expect_warning(throw_msg_if_duplicates(smetrics_data, "Metrics", msg_f = warning), exp_msg)
+  expect_error(throw_msg_if_duplicates(smetrics_dup_data, "Metrics"), exp_msg)
+  expect_warning(throw_msg_if_duplicates(smetrics_dup_data, "Metrics", msg_f = warning), exp_msg)
 })