add more types to dummy data

scverse · Sep 18, 2023 · 6169cc3 · 6169cc3
1 parent fc08613
commit 6169cc3
Showing 1 changed file with 52 additions and 16 deletions.
diff --git a/tests/testthat/helper-dummy_data.R b/tests/testthat/helper-dummy_data.R
@@ -11,11 +11,12 @@
 #'
 #' @examples
 #' dummy <- dummy_data()
-dummy_data <- function(n_obs = 10L, n_vars = 20L,
-                       output = c(
-                         "list", "SingleCellExperiment",
-                         "Seurat"
-                       )) {
+dummy_data <- function(
+    n_obs = 10L,
+    n_vars = 20L,
+    output = c(
+      "list", "SingleCellExperiment", "Seurat"
+    )) {
   output <- match.arg(output)
 
   switch(output,
@@ -39,21 +40,55 @@ dummy_list <- function(n_obs = 10L, n_vars = 20L) {
   # generate X
   X <- Matrix::rsparsematrix(nrow = n_obs, ncol = n_vars, density = .1)
 
-  # generate layers
+  # generate layers in dense, sparse, row compressed and column compressed
   layers <- list(
-    X2 = X * 2,
-    X3 = X * 3
+    dense = as.matrix(X),
+    CsparseMatrix = as(X, "CsparseMatrix"),
+    RsparseMatrix = as(X, "RsparseMatrix")
   )
 
-  # generate obs
+  # generate obs with different types (character, integer, factors, etc.)
   obs <- data.frame(
-    cell_type = sample(c("tcell", "bcell"), n_obs, replace = TRUE),
-    cluster = sample.int(3, n_obs, replace = TRUE)
+    character = paste0("cell", seq_len(n_obs)),
+    integer = seq_len(n_obs),
+    factor = factor(paste0("cell", seq_len(n_obs))),
+    factor_ordered = factor(paste0("cell", seq_len(n_obs)), ordered = TRUE),
+    logical = sample(c(TRUE, FALSE), n_obs, replace = TRUE),
+    numeric = runif(n_obs),
+    character_with_nas = c(paste0("cell", seq_len(n_obs - 1)), NA),
+    integer_with_nas = c(seq_len(n_obs - 1), NA),
+    factor_with_nas = c(
+      factor(paste0("cell", seq_len(n_obs - 1))),
+      NA_character_
+    ),
+    factor_ordered_with_nas = c(
+      factor(paste0("cell", seq_len(n_obs - 1)), ordered = TRUE),
+      NA_character_
+    ),
+    logical_with_nas = c(sample(c(TRUE, FALSE), n_obs - 1, replace = TRUE), NA),
+    numeric_with_nas = c(runif(n_obs - 1), NA)
   )
 
-  # generate var
+  # generate var with different types (character, integer, factors, etc.)
   var <- data.frame(
-    geneinfo = sample(c("a", "b", "c"), n_vars, replace = TRUE)
+    character = paste0("gene", seq_len(n_vars)),
+    integer = seq_len(n_vars),
+    factor = factor(paste0("gene", seq_len(n_vars))),
+    factor_ordered = factor(paste0("gene", seq_len(n_vars)), ordered = TRUE),
+    logical = sample(c(TRUE, FALSE), n_vars, replace = TRUE),
+    numeric = runif(n_vars),
+    character_with_nas = c(paste0("gene", seq_len(n_vars - 1)), NA),
+    integer_with_nas = c(seq_len(n_vars - 1), NA),
+    factor_with_nas = c(
+      factor(paste0("gene", seq_len(n_vars - 1))),
+      NA_character_
+    ),
+    factor_ordered_with_nas = c(
+      factor(paste0("gene", seq_len(n_vars - 1)), ordered = TRUE),
+      NA_character_
+    ),
+    logical_with_nas = c(sample(c(TRUE, FALSE), n_vars - 1, replace = TRUE), NA),
+    numeric_with_nas = c(runif(n_vars - 1), NA)
   )
 
   # generate obs_names
@@ -122,18 +157,19 @@ dummy_Seurat <- function(...) { # nolint
 
   dummy <- dummy_data(...)
 
-  X <- t(dummy$X)
+  X <- t(dummy$layers[["dense"]])
   colnames(X) <- dummy$obs_names
   rownames(X) <- dummy$var_names
 
   seurat <- SeuratObject::CreateSeuratObject(X)
 
-  X2 <- t(dummy$layers$X2)
+  X2 <- as(t(dummy$layers[["CsparseMatrix"]]), "CsparseMatrix")
   colnames(X2) <- dummy$obs_names
   rownames(X2) <- dummy$var_names
   seurat <- SeuratObject::SetAssayData(seurat, "data", X2)
 
-  X3 <- as.matrix(t(dummy$layers$X3))
+  # seurat doesn't support RsparseMatrices
+  X3 <- as.matrix(t(dummy$layers[["RsparseMatrix"]]))
   colnames(X3) <- dummy$obs_names
   rownames(X3) <- dummy$var_names
   seurat <- SeuratObject::SetAssayData(seurat, "scale.data", X3)