From e63726700028bce0aa29041be77870e64556bd68 Mon Sep 17 00:00:00 2001
From: ptajvar <ptajvar@gmail.com>
Date: Wed, 17 Jun 2026 14:13:42 +0000
Subject: [PATCH 01/11] allow fewer cell counts and markers in downsample

---
 R/read_data.R          | 98 ++++++++++++++++++++++++++++++++++++------
 man/downsample_data.Rd |  9 +++-
 2 files changed, 93 insertions(+), 14 deletions(-)

diff --git a/R/read_data.R b/R/read_data.R
index f9f7d45..15da5a7 100644
--- a/R/read_data.R
+++ b/R/read_data.R
@@ -277,14 +277,19 @@ merge_data <-
 
 #' Downsample data to a specified number of cells and markers
 #'
-#' Downsamples the data to a specified number of cells and markers, ensuring that control markers are always included.
+#' Downsamples the data to a specified number of cells and markers,
+#' ensuring that control markers are always included.
+#' If a sample has fewer than `n_cells`, all available cells are kept for that sample.
+#' If fewer non-control markers are available than requested, all available non-control markers are kept.
+#' If `length(control_markers) > n_markers`, all control markers are kept.
+#' In these cases, the function warns and proceeds instead of failing.
 #'
 #' @param pg_data A Seurat object containing the data to be downsampled.
 #' @param control_markers A character vector of control markers to always include in the downsampled data.
 #' @param n_cells An integer specifying the number of cells to keep in each sample.
 #' @param n_markers An integer specifying the total number of markers to keep in the downsampled data.
 #'
-#' @return A downsampled Seurat object with the specified number of cells and markers.
+#' @return A downsampled Seurat object with selected cells and markers.
 #'
 #' @export
 #'
@@ -295,24 +300,93 @@ downsample_data <-
            n_markers = 20) {
     set.seed(37)
 
-    keep_cells <-
+    if (is.null(control_markers)) {
+      control_markers <- character(0)
+    }
+
+    control_markers <- unique(control_markers)
+
+    cell_data <-
       FetchData(pg_data, "sample_alias") %>%
-      as_tibble(rownames = "cell_id") %>%
+      as_tibble(rownames = "cell_id")
+
+    available_cells <-
+      cell_data %>%
+      count(sample_alias, name = "n_available")
+
+    low_cell_samples <-
+      available_cells %>%
+      filter(n_available < n_cells)
+
+    if (nrow(low_cell_samples) > 0) {
+      warning(
+        paste0(
+          "Requested ", n_cells,
+          " cells per sample, but some samples have fewer cells. ",
+          "Using all available cells for those samples."
+        ),
+        call. = FALSE
+      )
+    }
+
+    keep_cells <-
+      cell_data %>%
       group_by(sample_alias) %>%
-      slice_sample(n = n_cells) %>%
+      group_modify(~ slice_sample(.x, n = min(nrow(.x), n_cells))) %>%
+      ungroup() %>%
       pull(cell_id)
 
-    pixelatorR:::assert_x_in_y(control_markers, rownames(pg_data))
+    if (length(control_markers) > 0) {
+      pixelatorR:::assert_x_in_y(control_markers, rownames(pg_data))
+    }
 
-    keep_markers <-
+    non_control_markers <-
       rownames(pg_data) %>%
       {
         .[!. %in% control_markers]
-      } %>%
-      {
-        .[sample(seq_along(.), size = n_markers - length(control_markers), replace = FALSE)]
-      } %>%
-      union(control_markers)
+      }
+
+    requested_non_control <- n_markers - length(control_markers)
+
+    if (requested_non_control <= 0) {
+      if (length(control_markers) > n_markers) {
+        warning(
+          paste0(
+            "Requested n_markers = ", n_markers,
+            " but control_markers has length ", length(control_markers),
+            ". Keeping all control markers."
+          ),
+          call. = FALSE
+        )
+      }
+      sampled_non_control <- character(0)
+    } else {
+      n_non_control <- min(length(non_control_markers), requested_non_control)
+
+      if (n_non_control < requested_non_control) {
+        warning(
+          paste0(
+            "Requested ", requested_non_control,
+            " non-control markers, but only ", n_non_control,
+            " are available. Using all available non-control markers."
+          ),
+          call. = FALSE
+        )
+      }
+
+      sampled_non_control <-
+        if (n_non_control > 0) {
+          non_control_markers[
+            sample(seq_along(non_control_markers), size = n_non_control, replace = FALSE)
+          ]
+        } else {
+          character(0)
+        }
+    }
+
+    keep_markers <-
+      c(control_markers, sampled_non_control) %>%
+      unique()
 
     pg_data <-
       pg_data[keep_markers, keep_cells]
diff --git a/man/downsample_data.Rd b/man/downsample_data.Rd
index 842dfaf..cfcc934 100644
--- a/man/downsample_data.Rd
+++ b/man/downsample_data.Rd
@@ -16,8 +16,13 @@ downsample_data(pg_data, control_markers = NULL, n_cells = 50, n_markers = 20)
 \item{n_markers}{An integer specifying the total number of markers to keep in the downsampled data.}
 }
 \value{
-A downsampled Seurat object with the specified number of cells and markers.
+A downsampled Seurat object with selected cells and markers.
 }
 \description{
-Downsamples the data to a specified number of cells and markers, ensuring that control markers are always included.
+Downsamples the data to a specified number of cells and markers,
+ensuring that control markers are always included.
+If a sample has fewer than \code{n_cells}, all available cells are kept for that sample.
+If fewer non-control markers are available than requested, all available non-control markers are kept.
+If \code{length(control_markers) > n_markers}, all control markers are kept.
+In these cases, the function warns and proceeds instead of failing.
 }

From d582e3a98cc503b708d48e32edc50c8d79428654 Mon Sep 17 00:00:00 2001
From: ptajvar <ptajvar@gmail.com>
Date: Wed, 17 Jun 2026 14:37:49 +0000
Subject: [PATCH 02/11] disable downsampling markers when a component's total
 count becomes zero.

---
 R/read_data.R          | 20 ++++++++++++++++++++
 man/downsample_data.Rd |  2 ++
 2 files changed, 22 insertions(+)

diff --git a/R/read_data.R b/R/read_data.R
index 15da5a7..29d6d8c 100644
--- a/R/read_data.R
+++ b/R/read_data.R
@@ -282,6 +282,8 @@ merge_data <-
 #' If a sample has fewer than `n_cells`, all available cells are kept for that sample.
 #' If fewer non-control markers are available than requested, all available non-control markers are kept.
 #' If `length(control_markers) > n_markers`, all control markers are kept.
+#' If marker downsampling makes any selected component have zero total counts,
+#' marker downsampling is disabled and all markers are kept.
 #' In these cases, the function warns and proceeds instead of failing.
 #'
 #' @param pg_data A Seurat object containing the data to be downsampled.
@@ -388,6 +390,24 @@ downsample_data <-
       c(control_markers, sampled_non_control) %>%
       unique()
 
+    candidate_data <-
+      pg_data[keep_markers, keep_cells]
+
+    marker_counts <-
+      GetAssayData(candidate_data, slot = "counts")
+
+    if (any(Matrix::colSums(marker_counts) == 0)) {
+      warning(
+        paste0(
+          "Marker downsampling produced components with zero total counts. ",
+          "Disabling marker downsampling and keeping all markers."
+        ),
+        call. = FALSE
+      )
+
+      keep_markers <- rownames(pg_data)
+    }
+
     pg_data <-
       pg_data[keep_markers, keep_cells]
 
diff --git a/man/downsample_data.Rd b/man/downsample_data.Rd
index cfcc934..1f85d78 100644
--- a/man/downsample_data.Rd
+++ b/man/downsample_data.Rd
@@ -24,5 +24,7 @@ ensuring that control markers are always included.
 If a sample has fewer than \code{n_cells}, all available cells are kept for that sample.
 If fewer non-control markers are available than requested, all available non-control markers are kept.
 If \code{length(control_markers) > n_markers}, all control markers are kept.
+If marker downsampling makes any selected component have zero total counts,
+marker downsampling is disabled and all markers are kept.
 In these cases, the function warns and proceeds instead of failing.
 }

From 8dbc57693f1467ed567f895dc448b1d274b5af39 Mon Sep 17 00:00:00 2001
From: Pouria Tajvar <31826342+ptajvar@users.noreply.github.com>
Date: Wed, 17 Jun 2026 16:56:05 +0200
Subject: [PATCH 03/11] Apply suggestions from code review

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
---
 R/read_data.R | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/R/read_data.R b/R/read_data.R
index 29d6d8c..e7d1528 100644
--- a/R/read_data.R
+++ b/R/read_data.R
@@ -282,7 +282,7 @@ merge_data <-
 #' If a sample has fewer than `n_cells`, all available cells are kept for that sample.
 #' If fewer non-control markers are available than requested, all available non-control markers are kept.
 #' If `length(control_markers) > n_markers`, all control markers are kept.
-#' If marker downsampling makes any selected component have zero total counts,
+#' If marker downsampling makes any selected cell have zero total counts,
 #' marker downsampling is disabled and all markers are kept.
 #' In these cases, the function warns and proceeds instead of failing.
 #'
@@ -399,7 +399,7 @@ downsample_data <-
     if (any(Matrix::colSums(marker_counts) == 0)) {
       warning(
         paste0(
-          "Marker downsampling produced components with zero total counts. ",
+          "Marker downsampling produced cells with zero total counts. ",
           "Disabling marker downsampling and keeping all markers."
         ),
         call. = FALSE

From c8bd0e8886b7fdfdd059965895bdaa473559e333 Mon Sep 17 00:00:00 2001
From: ptajvar <ptajvar@gmail.com>
Date: Wed, 17 Jun 2026 15:12:58 +0000
Subject: [PATCH 04/11] adding tests

---
 tests/testthat/test_read_data.R | 60 +++++++++++++++++++++++++++++++++
 1 file changed, 60 insertions(+)

diff --git a/tests/testthat/test_read_data.R b/tests/testthat/test_read_data.R
index 3359098..813f8d8 100644
--- a/tests/testthat/test_read_data.R
+++ b/tests/testthat/test_read_data.R
@@ -209,6 +209,66 @@ test_that("File reading works as expected", {
   expect_s4_class(seur_down, "Seurat")
   expect_equal(dim(seur_down), c(5, 6))
 
+  # Edge case 1: fewer cells than requested in at least one sample
+  expect_warning(
+    seur_down_low_cells <- downsample_data(
+      seur_comb,
+      control_markers = c("mIgG1", "mIgG2a", "mIgG2b"),
+      n_cells = 1000,
+      n_markers = 5
+    ),
+    "fewer cells"
+  )
+  expect_equal(ncol(seur_down_low_cells), ncol(seur_comb))
+
+  # Edge case 2: fewer non-control markers available than requested
+  all_markers <- rownames(seur_comb)
+  expect_gt(length(all_markers), 1)
+  control_set <- all_markers[-length(all_markers)]
+  expect_warning(
+    seur_down_low_markers <- downsample_data(
+      seur_comb,
+      control_markers = control_set,
+      n_cells = 3,
+      n_markers = nrow(seur_comb) + 5
+    ),
+    "non-control markers"
+  )
+  expect_equal(nrow(seur_down_low_markers), nrow(seur_comb))
+
+  # Edge case 3: marker downsampling would produce zero-total-count components,
+  # so all markers should be kept.
+  sparse_counts <-
+    matrix(
+      c(
+        1, 1, 0, 0,
+        0, 0, 1, 0,
+        0, 0, 0, 2
+      ),
+      nrow = 3,
+      byrow = TRUE,
+      dimnames = list(c("m1", "m2", "m3"), c("c1", "c2", "c3", "c4"))
+    )
+  sparse_seur <- SeuratObject::CreateSeuratObject(counts = sparse_counts)
+  sparse_seur <- SeuratObject::AddMetaData(
+    sparse_seur,
+    metadata = data.frame(
+      sample_alias = c("S1", "S1", "S2", "S2"),
+      row.names = colnames(sparse_seur)
+    )
+  )
+
+  expect_warning(
+    sparse_down <- downsample_data(
+      sparse_seur,
+      control_markers = c("m1", "m2"),
+      n_cells = 2,
+      n_markers = 2
+    ),
+    "zero total counts"
+  )
+  expect_equal(nrow(sparse_down), nrow(sparse_seur))
+
   # Sample sheet reading
   expect_no_error(sample_sheet <- read_samplesheet(test_samplesheet()))
   expect_equal(

From 8639d5aec8f8a054908926ccd28002a788f80f07 Mon Sep 17 00:00:00 2001
From: ptajvar <ptajvar@gmail.com>
Date: Wed, 17 Jun 2026 15:13:15 +0000
Subject: [PATCH 05/11] fixing documentation

---
 man/downsample_data.Rd | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/man/downsample_data.Rd b/man/downsample_data.Rd
index 1f85d78..fc69ab2 100644
--- a/man/downsample_data.Rd
+++ b/man/downsample_data.Rd
@@ -24,7 +24,7 @@ ensuring that control markers are always included.
 If a sample has fewer than \code{n_cells}, all available cells are kept for that sample.
 If fewer non-control markers are available than requested, all available non-control markers are kept.
 If \code{length(control_markers) > n_markers}, all control markers are kept.
-If marker downsampling makes any selected component have zero total counts,
+If marker downsampling makes any selected cell have zero total counts,
 marker downsampling is disabled and all markers are kept.
 In these cases, the function warns and proceeds instead of failing.
 }

From 9ef620cb0e64a3da576a633711570f2e18834f8e Mon Sep 17 00:00:00 2001
From: ptajvar <ptajvar@gmail.com>
Date: Wed, 17 Jun 2026 15:13:51 +0000
Subject: [PATCH 06/11] use layer instead of deprecated slot in GetAssayData

---
 R/read_data.R | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/R/read_data.R b/R/read_data.R
index e7d1528..c56f8d2 100644
--- a/R/read_data.R
+++ b/R/read_data.R
@@ -334,7 +334,7 @@ downsample_data <-
     keep_cells <-
       cell_data %>%
       group_by(sample_alias) %>%
-      group_modify(~ slice_sample(.x, n = min(nrow(.x), n_cells))) %>%
+      slice_sample(n = min(dplyr::n(), n_cells)) %>%
       ungroup() %>%
       pull(cell_id)
 
@@ -394,7 +394,7 @@ downsample_data <-
       pg_data[keep_markers, keep_cells]
 
     marker_counts <-
-      GetAssayData(candidate_data, slot = "counts")
+          GetAssayData(candidate_data, layer = "counts")
 
     if (any(Matrix::colSums(marker_counts) == 0)) {
       warning(

From 440d8db8cea3615bba71317ba85d03578db53682 Mon Sep 17 00:00:00 2001
From: ptajvar <ptajvar@gmail.com>
Date: Wed, 17 Jun 2026 15:22:14 +0000
Subject: [PATCH 07/11] applying comments from the PR review.

---
 R/read_data.R | 83 ++++++++++++++++++++++-----------------------------
 1 file changed, 36 insertions(+), 47 deletions(-)

diff --git a/R/read_data.R b/R/read_data.R
index c56f8d2..627f14f 100644
--- a/R/read_data.R
+++ b/R/read_data.R
@@ -321,13 +321,11 @@ downsample_data <-
       filter(n_available < n_cells)
 
     if (nrow(low_cell_samples) > 0) {
-      warning(
-        paste0(
-          "Requested ", n_cells,
-          " cells per sample, but some samples have fewer cells. ",
-          "Using all available cells for those samples."
-        ),
-        call. = FALSE
+      cli::cli_warn(
+        c(
+          "Requested {.val {n_cells}} cells per sample, but some samples have fewer cells.",
+          "i" = "Using all available cells for those samples."
+        )
       )
     }
 
@@ -348,44 +346,36 @@ downsample_data <-
         .[!. %in% control_markers]
       }
 
-    requested_non_control <- n_markers - length(control_markers)
-
-    if (requested_non_control <= 0) {
-      if (length(control_markers) > n_markers) {
-        warning(
-          paste0(
-            "Requested n_markers = ", n_markers,
-            " but control_markers has length ", length(control_markers),
-            ". Keeping all control markers."
-          ),
-          call. = FALSE
-        )
-      }
-      sampled_non_control <- character(0)
-    } else {
-      n_non_control <- min(length(non_control_markers), requested_non_control)
-
-      if (n_non_control < requested_non_control) {
-        warning(
-          paste0(
-            "Requested ", requested_non_control,
-            " non-control markers, but only ", n_non_control,
-            " are available. Using all available non-control markers."
-          ),
-          call. = FALSE
+    target_non_control <- max(n_markers - length(control_markers), 0)
+    available_non_control <- length(non_control_markers)
+
+    if (length(control_markers) > n_markers) {
+      cli::cli_warn(
+        c(
+          "Requested {.val {n_markers}} total markers, but {.val {length(control_markers)}} control markers were provided.",
+          "i" = "Keeping all control markers."
         )
-      }
+      )
+    }
 
-      sampled_non_control <-
-        if (n_non_control > 0) {
-          non_control_markers[
-            sample(seq_along(non_control_markers), size = n_non_control, replace = FALSE)
-          ]
-        } else {
-          character(0)
-        }
+    if (target_non_control > available_non_control) {
+      cli::cli_warn(
+        c(
+          "Requested {.val {target_non_control}} non-control markers, but only {.val {available_non_control}} are available.",
+          "i" = "Using all available non-control markers."
+        )
+      )
     }
 
+    n_non_control <- min(target_non_control, available_non_control)
+
+    sampled_non_control <-
+      if (n_non_control > 0) {
+        sample(non_control_markers, size = n_non_control, replace = FALSE)
+      } else {
+        character(0)
+      }
+
     keep_markers <-
       c(control_markers, sampled_non_control) %>%
       unique()
@@ -397,12 +387,11 @@ downsample_data <-
           GetAssayData(candidate_data, layer = "counts")
 
     if (any(Matrix::colSums(marker_counts) == 0)) {
-      warning(
-        paste0(
-          "Marker downsampling produced cells with zero total counts. ",
-          "Disabling marker downsampling and keeping all markers."
-        ),
-        call. = FALSE
+      cli::cli_warn(
+        c(
+          "Marker downsampling produced cells with zero total counts.",
+          "i" = "Disabling marker downsampling and keeping all markers."
+        )
       )
 
       keep_markers <- rownames(pg_data)

From 40f1a4aefe5588c444850b8fa9bad1f5ebf520a2 Mon Sep 17 00:00:00 2001
From: ptajvar <ptajvar@gmail.com>
Date: Wed, 17 Jun 2026 15:25:15 +0000
Subject: [PATCH 08/11] fix dplyr import

---
 R/read_data.R | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/R/read_data.R b/R/read_data.R
index 627f14f..2c338fa 100644
--- a/R/read_data.R
+++ b/R/read_data.R
@@ -332,7 +332,7 @@ downsample_data <-
     keep_cells <-
       cell_data %>%
       group_by(sample_alias) %>%
-      slice_sample(n = min(dplyr::n(), n_cells)) %>%
+      dplyr::group_modify(~ slice_sample(.x, n = min(nrow(.x), n_cells))) %>%
       ungroup() %>%
       pull(cell_id)
 

From 9416d0ace6a6f33584d9e64682bfbe966fde0016 Mon Sep 17 00:00:00 2001
From: ptajvar <ptajvar@gmail.com>
Date: Wed, 17 Jun 2026 15:30:57 +0000
Subject: [PATCH 09/11] simplify and fix linting.

---
 R/read_data.R | 11 ++---------
 1 file changed, 2 insertions(+), 9 deletions(-)

diff --git a/R/read_data.R b/R/read_data.R
index 2c338fa..de130e6 100644
--- a/R/read_data.R
+++ b/R/read_data.R
@@ -349,19 +349,12 @@ downsample_data <-
     target_non_control <- max(n_markers - length(control_markers), 0)
     available_non_control <- length(non_control_markers)
 
-    if (length(control_markers) > n_markers) {
-      cli::cli_warn(
-        c(
-          "Requested {.val {n_markers}} total markers, but {.val {length(control_markers)}} control markers were provided.",
-          "i" = "Keeping all control markers."
-        )
-      )
-    }
 
     if (target_non_control > available_non_control) {
       cli::cli_warn(
         c(
-          "Requested {.val {target_non_control}} non-control markers, but only {.val {available_non_control}} are available.",
+          "Requested {.val {target_non_control}} non-control markers,",
+          "but only {.val {available_non_control}} are available.",
           "i" = "Using all available non-control markers."
         )
       )

From 6a974cbf148fa486710e8806ed6c035a54cfa76b Mon Sep 17 00:00:00 2001
From: maxkarlsson <max.karlsson@pixelgen.com>
Date: Thu, 18 Jun 2026 09:43:56 +0200
Subject: [PATCH 10/11] fix: final touches

---
 R/read_data.R                   | 41 ++++++++++-----------------------
 man/downsample_data.Rd          |  4 +---
 tests/testthat/test_read_data.R | 33 --------------------------
 3 files changed, 13 insertions(+), 65 deletions(-)

diff --git a/R/read_data.R b/R/read_data.R
index de130e6..9ff1fe1 100644
--- a/R/read_data.R
+++ b/R/read_data.R
@@ -281,9 +281,7 @@ merge_data <-
 #' ensuring that control markers are always included.
 #' If a sample has fewer than `n_cells`, all available cells are kept for that sample.
 #' If fewer non-control markers are available than requested, all available non-control markers are kept.
-#' If `length(control_markers) > n_markers`, all control markers are kept.
-#' If marker downsampling makes any selected cell have zero total counts,
-#' marker downsampling is disabled and all markers are kept.
+#' `control_markers` are always kept.
 #' In these cases, the function warns and proceeds instead of failing.
 #'
 #' @param pg_data A Seurat object containing the data to be downsampled.
@@ -302,12 +300,18 @@ downsample_data <-
            n_markers = 20) {
     set.seed(37)
 
-    if (is.null(control_markers)) {
-      control_markers <- character(0)
+    pixelatorR:::assert_class(pg_data, "Seurat")
+    pixelatorR:::assert_vector(control_markers, "character", allow_null = TRUE)
+    pixelatorR:::assert_single_value(n_cells, "integer")
+    pixelatorR:::assert_single_value(n_markers, "integer")
+
+    if (length(control_markers) > 0) {
+      pixelatorR:::assert_x_in_y(control_markers, rownames(pg_data))
     }
 
     control_markers <- unique(control_markers)
 
+    # Downsample cells
     cell_data <-
       FetchData(pg_data, "sample_alias") %>%
       as_tibble(rownames = "cell_id")
@@ -336,15 +340,11 @@ downsample_data <-
       ungroup() %>%
       pull(cell_id)
 
-    if (length(control_markers) > 0) {
-      pixelatorR:::assert_x_in_y(control_markers, rownames(pg_data))
-    }
 
+    # Downsample markers
+    markers <- rownames(pg_data)
     non_control_markers <-
-      rownames(pg_data) %>%
-      {
-        .[!. %in% control_markers]
-      }
+      setdiff(markers, control_markers)
 
     target_non_control <- max(n_markers - length(control_markers), 0)
     available_non_control <- length(non_control_markers)
@@ -373,23 +373,6 @@ downsample_data <-
       c(control_markers, sampled_non_control) %>%
       unique()
 
-    candidate_data <-
-      pg_data[keep_markers, keep_cells]
-
-    marker_counts <-
-          GetAssayData(candidate_data, layer = "counts")
-
-    if (any(Matrix::colSums(marker_counts) == 0)) {
-      cli::cli_warn(
-        c(
-          "Marker downsampling produced cells with zero total counts.",
-          "i" = "Disabling marker downsampling and keeping all markers."
-        )
-      )
-
-      keep_markers <- rownames(pg_data)
-    }
-
     pg_data <-
       pg_data[keep_markers, keep_cells]
 
diff --git a/man/downsample_data.Rd b/man/downsample_data.Rd
index fc69ab2..99c8b8f 100644
--- a/man/downsample_data.Rd
+++ b/man/downsample_data.Rd
@@ -23,8 +23,6 @@ Downsamples the data to a specified number of cells and markers,
 ensuring that control markers are always included.
 If a sample has fewer than \code{n_cells}, all available cells are kept for that sample.
 If fewer non-control markers are available than requested, all available non-control markers are kept.
-If \code{length(control_markers) > n_markers}, all control markers are kept.
-If marker downsampling makes any selected cell have zero total counts,
-marker downsampling is disabled and all markers are kept.
+\code{control_markers} are always kept.
 In these cases, the function warns and proceeds instead of failing.
 }
diff --git a/tests/testthat/test_read_data.R b/tests/testthat/test_read_data.R
index 813f8d8..5ae3816 100644
--- a/tests/testthat/test_read_data.R
+++ b/tests/testthat/test_read_data.R
@@ -236,39 +236,6 @@ test_that("File reading works as expected", {
   )
   expect_equal(nrow(seur_down_low_markers), nrow(seur_comb))
 
-  # Edge case 3: marker downsampling would produce zero-total-count components,
-  # so all markers should be kept.
-  sparse_counts <-
-    matrix(
-      c(
-        1, 1, 0, 0,
-        0, 0, 1, 0,
-        0, 0, 0, 2
-      ),
-      nrow = 3,
-      byrow = TRUE,
-      dimnames = list(c("m1", "m2", "m3"), c("c1", "c2", "c3", "c4"))
-    )
-  sparse_seur <- SeuratObject::CreateSeuratObject(counts = sparse_counts)
-  sparse_seur <- SeuratObject::AddMetaData(
-    sparse_seur,
-    metadata = data.frame(
-      sample_alias = c("S1", "S1", "S2", "S2"),
-      row.names = colnames(sparse_seur)
-    )
-  )
-
-  expect_warning(
-    sparse_down <- downsample_data(
-      sparse_seur,
-      control_markers = c("m1", "m2"),
-      n_cells = 2,
-      n_markers = 2
-    ),
-    "zero total counts"
-  )
-  expect_equal(nrow(sparse_down), nrow(sparse_seur))
-
   # Sample sheet reading
   expect_no_error(sample_sheet <- read_samplesheet(test_samplesheet()))
   expect_equal(

From 92e8ccdc13af30e600208d39ed4cc3ea3f925e1d Mon Sep 17 00:00:00 2001
From: maxkarlsson <max.karlsson@pixelgen.com>
Date: Thu, 18 Jun 2026 14:50:21 +0200
Subject: [PATCH 11/11] chore: changelog

---
 CHANGELOG.md | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index cdfba55..d16a87f 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -5,13 +5,19 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [Unreleased]
+
+### Fixed
+ 
+ - `downsample_data` will now default to the number of markers or cells in the data if `n_markers` or `n_cells` are higher than available in the data.
+
 ## [0.11.2] 2026-06-15
 
-### Changes 
+### Changed
 
  - `component_hashing` now returns a sample confidence plot with either hash purity or hash enrichment factor, depending on which metric is present in the data.
  
- ### Removed
+### Removed
  
  - `harmony` has been removed, and `do_harmonize` is no longer an option.