Merge pull request #361 from hms-dbmi-cellenics/fix-factor-active-ident

make rename_active_ident work with factor+character

pipeline-runner/R/seurat-3-upload_seurat_to_aws.R

@@ -83,10 +83,10 @@ upload_seurat_to_aws <- function(input, pipeline_config, prev_out) {
 rename_active_ident <- function(meta) {
 
   # determine columns that are duplicates of active.ident
-  test_vals <- meta$active.ident
+  test_vals <- as.character(meta$active.ident)
   dup.cols <- c()
   for (col in setdiff(colnames(meta), 'active.ident')) {
-    col_vals <- meta[[col]]
+    col_vals <- as.character(meta[[col]])
     if (identical(test_vals, col_vals))
       dup.cols <- c(dup.cols, col)
   }

pipeline-runner/tests/testthat/test-seurat-3-upload_seurat_to_aws.R

@@ -344,3 +344,21 @@ test_that("rename_active_ident does not rename active.ident when there isn't a d
   meta_renamed <- rename_active_ident(meta)
   expect_identical(meta, meta_renamed)
 })
+
+test_that("rename_active_ident works when one column is factor and the other is character", {
+
+  meta <- data.frame(
+    active.ident = c('cluster1', 'cluster2', 'cluster3'),
+    cell_type = factor(c('cluster1', 'cluster2', 'cluster3')),
+    other_clustering = c('clusterA', 'clusterA', 'clusterB')
+  )
+
+  expected_name <- 'cell_type'
+
+  meta_renamed <- rename_active_ident(meta)
+
+  expect_equal(colnames(meta_renamed)[1], expected_name)
+  expect_false('active.ident' %in% colnames(meta_renamed))
+  expect_equal(colnames(meta_renamed), c('cell_type', 'other_clustering'))
+})
+