wrap_dataset(): Repurpose function to be able to choose the output …

…format using a parameter (#28).

NAMESPACE

@@ -3,6 +3,7 @@
 export("%>%")
 export(as_anndata)
 export(as_dyno)
+export(as_list)
 export(as_sce)
 export(as_seurat)
 export(backbone)

NEWS.md

@@ -1,3 +1,11 @@
+# dyngen 1.0.1
+
+## NEW FEATURES
+
+* `wrap_dataset()`: Repurpose function to be able to choose the output format using a parameter (#28).
+
+* `as_list()`: Added a function for converting the dyngen output to a simple list object.
+
 # dyngen 1.0.0
 
 This version mostly upgrades dyngen's ease-of-use, such as better vignettes, conversion functions for working with dyngen datasets in other packages, and more useful ways of specifying platform-specific parameters (i.e. number of cores and cache location). Perhaps more excitingly, the dyngen documentation is more readable online at [https://dyngen.dynverse.org](https://dyngen.dynverse.org)!

R/8_convert.R

@@ -1,6 +1,6 @@
 #' Convert simulation output to different formats.
 #' 
-#' For use with other packages compatible with dyno / anndata.
+#' For use with other packages compatible with dyno, anndata, SingleCellExperiment, or Seurat.
 #' 
 #' @param model A dyngen output model for which the experiment has been emulated with [generate_experiment()].
 #' @param store_cellwise_grn Whether or not to also store cellwise GRN information.
@@ -131,23 +131,6 @@ as_dyno <- function(
   dataset
 }
 
-#' @rdname convert
-#' @export
-wrap_dataset <- function(
-  model,
-  store_dimred = !is.null(model$simulations$dimred),
-  store_cellwise_grn = !is.null(model$experiment$cellwise_grn),
-  store_rna_velocity = !is.null(model$experiment$rna_velocity)
-) {
-  .Deprecated("as_dyno")
-  as_dyno(
-    model = model, 
-    store_dimred = store_dimred,
-    store_cellwise_grn = store_cellwise_grn,
-    store_rna_velocity = store_rna_velocity
-  )
-}
-
 
 #' @rdname convert
 #' @importFrom tibble column_to_rownames
@@ -386,4 +369,107 @@ as_seurat <- function(
   }
 
   seurat_obj
+}
+
+#' @rdname convert
+#' @export
+as_list <- function(
+  model,
+  store_dimred = !is.null(model$simulations$dimred),
+  store_cellwise_grn = !is.null(model$experiment$cellwise_grn),
+  store_rna_velocity = !is.null(model$experiment$rna_velocity)
+) {
+  assert_that(
+    !is.null(model$experiment), 
+    msg = "model should be an object that was initialised with `initialise_model()`."
+  )
+
+  counts <- model$experiment$counts_mrna + model$experiment$counts_premrna
+  dataset <- list(
+    id = model$id,
+    cell_ids = rownames(counts),
+    feature_ids = colnames(counts),
+    counts = counts,
+    counts_spliced = model$experiment$counts_mrna,
+    counts_unspliced = model$experiment$counts_premrna,
+    counts_protein = model$experiment$counts_protein,
+    expression = as(log2(counts + 1), "dgCMatrix"),
+    cell_info = model$experiment$cell_info %>% select(-.data$from, -.data$to, -.data$time),
+    feature_info = model$experiment$feature_info,
+    milestone_ids = unique(c(model$gold_standard$network$from, model$gold_standard$network$to)),
+    milestone_network = model$gold_standard$network,
+    progressions = model$experiment$cell_info %>%
+      select(.data$cell_id, .data$from, .data$to, percentage = .data$time)
+  )
+
+  if (store_dimred) {
+    dimred <- model$simulations$dimred[model$experiment$cell_info$step_ix, , drop = FALSE]
+    rownames(dimred) <- model$experiment$cell_info$cell_id
+
+    dataset$dimred <- dimred
+    dataset$dimred_segment_points <- model$gold_standard$dimred[!model$gold_standard$meta$burn, , drop = FALSE]
+    dataset$dimred_segment_progressions <- model$gold_standard$meta[!model$gold_standard$meta$burn, , drop = FALSE] %>% 
+      select(.data$from, .data$to, percentage = .data$time)
+  }
+
+  # add a few more values
+  if (store_cellwise_grn) {
+    regulatory_network <- model$feature_network %>% 
+      select(regulator = .data$from, target = .data$to, .data$strength, .data$effect)
+    regulation_sc <- model$experiment$cellwise_grn
+
+    regulators <- unique(regulatory_network$regulator)
+    targets <- colnames(dataset$counts)
+    regulatory_network_sc <- 
+      Matrix::summary(regulation_sc) %>% 
+      transmute(
+        cell_id = factor(dataset$cell_ids[.data$i], levels = dataset$cell_ids),
+        regulator = factor(regulatory_network$regulator[.data$j], levels = regulators),
+        target = factor(regulatory_network$target[.data$j], levels = targets),
+        strength = .data$x
+      ) %>% 
+      as_tibble()
+
+
+    dataset$regulatory_network <- regulatory_network
+    dataset$regulatory_network_sc <- regulatory_network_sc
+    dataset$regulators <- regulators
+    dataset$targets <- targets
+  }
+
+  if (store_rna_velocity) {
+    dataset$rna_velocity <- model$experiment$rna_velocity
+  }
+
+  dataset
+}
+
+conversion_funs <- list(
+  dyno = as_dyno,
+  sce = as_sce,
+  seurat = as_seurat,
+  anndata = as_anndata,
+  list = as_list,
+  none = function(...) NULL
+)
+
+#' @rdname convert
+#' @param format Which output format to use, must be one of 'dyno', 'sce', 'seurat', 'anndata', 'list' or 'none'.
+#' @export
+wrap_dataset <- function(
+  model,
+  format = c("dyno", "sce", "seurat", "anndata", "list", "none"),
+  store_dimred = !is.null(model$simulations$dimred),
+  store_cellwise_grn = !is.null(model$experiment$cellwise_grn),
+  store_rna_velocity = !is.null(model$experiment$rna_velocity)
+) {
+  format <- match.arg(format, choices = c("dyno", "sce", "seurat", "anndata", "list", "none"))
+  fun <- conversion_funs[[format]]
+
+  fun(
+    model = model, 
+    store_dimred = store_dimred,
+    store_cellwise_grn = store_cellwise_grn,
+    store_rna_velocity = store_rna_velocity
+  )
 }

R/generate_dataset.R

@@ -8,10 +8,8 @@
 #' @param output_dir If not `NULL`, then the generated model and dynwrap 
 #'   dataset will be written to files in this directory.
 #' @param make_plots Whether or not to generate an overview of the dataset.
-#' @param dataset_format Which format to store the output dataset in.
-#'   Must be one of "dyno", "sce", "seurat", "anndata" or "none".
 #' 
-#' @inheritParams as_dyno
+#' @inheritParams wrap_dataset
 #' 
 #' @importFrom patchwork wrap_plots plot_annotation plot_layout
 #' @importFrom ggplot2 ggsave
@@ -54,16 +52,16 @@
 #' # dynplot::plot_dimred(dataset)
 #' }
 generate_dataset <- function(
-  model, 
+  model,
+  format = c("dyno", "sce", "seurat", "anndata", "list", "none"), 
   output_dir = NULL,
   make_plots = FALSE, 
   store_dimred = model$simulation_params$compute_dimred,
   store_cellwise_grn = model$simulation_params$compute_cellwise_grn,
-  store_rna_velocity = model$simulation_params$compute_rna_velocity,
-  dataset_format = c("dyno", "sce", "seurat", "anndata", "none")
+  store_rna_velocity = model$simulation_params$compute_rna_velocity
 ) {
   assert_that(is(model, "dyngen::init"))
-  dataset_format <- match.arg(dataset_format)
+  format <- match.arg(format, choices = c("dyno", "sce", "seurat", "anndata", "list", "none"))
 
   model <- model %>% 
     generate_tf_network() %>% 
@@ -73,23 +71,14 @@ generate_dataset <- function(
     generate_cells() %>% 
     generate_experiment()
 
-  dataset_function <- 
-    case_when(
-      dataset_format == "dyno" ~ as_dyno,
-      dataset_format == "sce" ~ as_sce,
-      dataset_format == "seurat" ~ as_seurat,
-      dataset_format == "anndata" ~ as_anndata,
-      dataset_format == "none" ~ NULL,
-      TRUE ~ stop("invalid dataset format, must be one of 'dyno', 'sce', 'seurat', 'anndata' or 'none'.")
-    )
-
-  if (dataset_format != "none") {
-    if (model$verbose) cat("Wrapping dataset as ", dataset_format, "\n", sep = "")
-    dataset <- dataset_function(
+  if (format != "none") {
+    if (model$verbose) cat("Wrapping dataset as ", format, "\n", sep = "")
+    dataset <- wrap_dataset(
       model, 
       store_dimred = store_dimred, 
       store_cellwise_grn = store_cellwise_grn,
-      store_rna_velocity = store_rna_velocity
+      store_rna_velocity = store_rna_velocity,
+      format = format
     )
   }
 
@@ -98,9 +87,9 @@ generate_dataset <- function(
     if (model$verbose) cat("Writing model to file\n")
     dir.create(dirname(output_dir), showWarnings = FALSE, recursive = FALSE)
 
-    if (dataset_format == "anndata") {
+    if (format == "anndata") {
       dataset$write_h5ad(paste0(output_dir, "dataset.h5ad"))
-    } else if (dataset_format != "none") {
+    } else if (format != "none") {
       saveRDS(dataset, paste0(output_dir, "dataset.rds"))
     }
 
@@ -139,10 +128,7 @@ generate_dataset <- function(
   }
 
   if (is.null(output_dir)) {
-    out <- list(model = model)
-    if (dataset_format != "none") {
-      out$dataset <- dataset
-    }
+    out <- list(model = model, dataset = dataset)
     if (make_plots) {
       out$plot <- g
     }