diff --git a/pipeline-runner/R/seurat-3-upload_seurat_to_aws.R b/pipeline-runner/R/seurat-3-upload_seurat_to_aws.R
index 9985f105..1e4c14d7 100644
--- a/pipeline-runner/R/seurat-3-upload_seurat_to_aws.R
+++ b/pipeline-runner/R/seurat-3-upload_seurat_to_aws.R
@@ -78,17 +78,18 @@ upload_seurat_to_aws <- function(input, pipeline_config, prev_out) {
 }
 
 find_cluster_columns <- function(scdata) {
+  meta <- scdata@meta.data
 
   # exclude all group columns, including duplicates
-  group_cols <- find_group_columns(scdata@meta.data, remove.dups = FALSE)
-  exclude_cols <- c(group_cols, 'samples')
+  group_cols <- find_group_columns(meta, remove.dups = FALSE)
+  group_cols <- c(group_cols, 'samples')
+  scdblfinder_cols <- grep('^scDblFinder', colnames(meta), value = TRUE)
 
   # order meta to indicate preference for louvain clusters
-  meta <- scdata@meta.data
   louvain_cols <- c('louvain', 'active.ident', 'seurat_clusters')
   meta <- meta |> dplyr::relocate(dplyr::any_of(louvain_cols))
 
-  check_cols <- setdiff(colnames(meta), exclude_cols)
+  check_cols <- setdiff(colnames(meta), c(scdblfinder_cols, group_cols))
 
   cluster_cols <- c()
   for (check_col in check_cols) {
@@ -111,9 +112,9 @@ find_cluster_columns <- function(scdata) {
 
     # skip if col is same as samples or group column
     is_sample_col <- FALSE
-    for (exclude_col in exclude_cols)  {
-      exclude_vals <- meta[[exclude_col]]
-      if (test_groups_equal(check_vals, exclude_vals)) {
+    for (group_col in group_cols)  {
+      group_vals <- meta[[group_col]]
+      if (test_groups_equal(check_vals, group_vals)) {
         is_sample_col <- TRUE
         break
       }
diff --git a/pipeline-runner/tests/testthat/test-seurat-3-upload_seurat_to_aws.R b/pipeline-runner/tests/testthat/test-seurat-3-upload_seurat_to_aws.R
index e8fcef7e..c88ead69 100644
--- a/pipeline-runner/tests/testthat/test-seurat-3-upload_seurat_to_aws.R
+++ b/pipeline-runner/tests/testthat/test-seurat-3-upload_seurat_to_aws.R
@@ -285,3 +285,19 @@ test_that("find_cluster_columns puts 'louvain' column first if exists", {
   cluster_cols <- find_cluster_columns(scdata)
   expect_equal(cluster_cols[1], 'louvain')
 })
+
+
+test_that("find_cluster_columns omits columns that start with scDblFinder", {
+
+  expected_cols <- c('RNA_snn_res.0.8', 'letter.idents', 'groups', 'RNA_snn_res.1')
+
+  scdata <- mock_scdata()
+  sample_names <- c('A', 'B', 'C', 'D')
+  samples <- rep(sample_names, each = ncol(scdata)/4)
+  scdata$samples <- samples
+
+  scdata$scDblFinder.cluster <- scdata$RNA_snn_res.0.8
+
+  cluster_cols <- find_cluster_columns(scdata)
+  expect_setequal(cluster_cols, expected_cols)
+})