CHOP|PROV|JHU only if checking for missiness in NAACCR code columns

Sage-Bionetworks · Sep 16, 2024 · 577b299 · 577b299
1 parent 65f371d
commit 577b299
Show file tree

Hide file tree

Showing 2 changed files with 27 additions and 36 deletions.
diff --git a/scripts/case_selection/shared_fxns.R b/scripts/case_selection/shared_fxns.R
@@ -294,15 +294,19 @@ remap_patient_characteristics <- function(clinical, existing_patients, ethnicity
 #' @param data The data frame to check against
 #' @param columns The target columns
 check_for_missing_values <- function(data, columns) {
-  # filter out CHOP, PROV, JHU centers with known NAs
-  data <- data[!grepl("CHOP|PROV|JHU", data$genie_patient_id), ]
   # Check for NA values or empty strings
   missingness_col <- c()
   for (col in columns) {
-    if (any(is.na(data[[col]]) | data[[col]] == "" )){
+    if (col %in% c("naaccr_race_code_tertiary", "naaccr_race_code_secondary")) {
+      # filter out CHOP, PROV, JHU centers with known NAs in NAACCR code columns
+      relevant_rows <- data[!grepl("CHOP|PROV|JHU", data$genie_patient_id), ]
+    } else{
+      relevant_rows <- data
+    }
+    if (any(is.na(relevant_rows[[col]]) | relevant_rows[[col]] == "" )){
       missingness_col <- c(col, missingness_col)
     }
-  }
+    }
   if (length(missingness_col) > 0) {
     warning(paste0("Warning: Missing or empty values found in column(s): ", paste(missingness_col,collapse=", ")))
     }

diff --git a/scripts/case_selection/tests/test_shared_fxns.R b/scripts/case_selection/tests/test_shared_fxns.R
@@ -104,51 +104,38 @@ test_that("remap_patient_characteristics works as expected", {
   expect_equal(result, expected_output)
 })
 
-test_that("check_for_missing_values - no missing or empty values in centers other than CHOP, PROV, JHU", {
+test_that("check_for_missing_values - no missing or empty values", {
   data <- data.frame(
-    col1 = c(1, 2, 3, NA),
-    col2 = c("a", "b", "c", ""),
-    genie_patient_id = c('a', 'b', 'c', 'CHOP123')
+    col1 = c(1, 2, 3),
+    col2 = c("a", "b", "c"),
+    genie_patient_id = c("a", "b", "CHOP123"),
+    naaccr_race_code_tertiary = c("a", "b", "c"),
+    naaccr_race_code_secondary = c("a", "b", "c")
   )
-  expect_warning(check_for_missing_values(data, c("col1", "col2")), NA)
+  expect_no_warning(check_for_missing_values(data, c("col1", "col2", "naaccr_race_code_tertiary", "naaccr_race_code_secondary")))
 
 })
 
-test_that("check_for_missing_values - NAs are detected in centers other than CHOP, PROV, JHU", {
+test_that("check_for_missing_values - missingness values are detected in NAACCR code columns in centers other than CHOP, PROV, JHU", {
   data <- data.frame(
-    col1 = c(1, NA, 3),
+    col1 = c(1, NA, ""),
     col2 = c("a", "b", "c"),
-    genie_patient_id = c('CHOP123', 'b', 'PROV234')
+    genie_patient_id = c("CHOP123", "b", "PROV234"),
+    naaccr_race_code_tertiary = c("a", "", "c"),
+    naaccr_race_code_secondary = c("a", "b", "c")
   )
-  expect_warning(check_for_missing_values(data, c("col1", "col2")), 
-              "Warning: Missing or empty values found in column\\(s\\): col1")
+  expect_warning(check_for_missing_values(data, c("col1", "col2", "naaccr_race_code_tertiary", "naaccr_race_code_secondary")), 
+              "Warning: Missing or empty values found in column\\(s\\): naaccr_race_code_tertiary, col1")
 })
 
-test_that("check_for_missing_values - empty string values are detected in centers other than CHOP, PROV, JHU", {
-  data <- data.frame(
-    col1 = c(1, 2, 3),
-    col2 = c("a", "", "c"),
-    genie_patient_id = c('CHOP123', 'b', 'PROV234')
-  )
-  expect_warning(check_for_missing_values(data, c("col1", "col2")), 
-               "Warning: Missing or empty values found in column\\(s\\): col2")
-})
-
-test_that("check_for_missing_values - multiple missing and empty values are detected in centers other than CHOP, PROV, JHU", {
+test_that("check_for_missing_values - missingness values are detected in NAACCR code columns in CHOP, PROV, JHU centers", {
   data <- data.frame(
     col1 = c(1, NA, ""),
     col2 = c("a", "", "c"),
-    genie_patient_id = c('CHOP123', 'b', 'PROV234')
+    genie_patient_id = c("CHOP123", "b", "PROV234"),
+    naaccr_race_code_tertiary = c("", "b", "c"),
+    naaccr_race_code_secondary = c("a", "b", NA)
   )
-  expect_warning(check_for_missing_values(data, c("col1", "col2")), 
+  expect_warning(check_for_missing_values(data, c("col1", "col2", "naaccr_race_code_tertiary", "naaccr_race_code_secondary")), 
                "Warning: Missing or empty values found in column\\(s\\): col2, col1")
 })
-
-test_that("check_for_missing_values - multiple missing and empty values are detected in CHOP, PROV, JHU centers", {
-  data <- data.frame(
-    col1 = c(1, NA, 2),
-    col2 = c("a", "", "c"),
-    genie_patient_id = c('a', 'CHOP123', 'PROV234')
-  )
-  expect_warning(check_for_missing_values(data, c("col1", "col2")), NA)
-})