Merge pull request #114 from biomage-org/patch-ensure-cell-ids-are-li…

…st-in-cell-sets

Patch, ensure `cellIds` is sent as an array always

pipeline-runner/R/gem2s-7-upload_to_aws.R

@@ -146,7 +146,7 @@ build_sample_cellsets <- function(input, scdata_list, color_pool) {
       key = sample_ids[i],
       name = sample_names[i],
       color = color_pool[i],
-      cellIds = unname(cell_ids)
+      cellIds = ensure_is_list_in_json(unname(cell_ids))
     )
   }
 
@@ -375,7 +375,7 @@ build_scratchpad_cellsets <- function(color_pool, subset_cellsets) {
       key = scratchpad_ids[i],
       name = scratchpad_names[i],
       color = color_pool[i],
-      cellIds = subset_cellsets[key == scratchpad_ids[i], cell_id]
+      cellIds =  ensure_is_list_in_json(subset_cellsets[key == scratchpad_ids[i], cell_id])
     )
   }
 

pipeline-runner/R/qc-7-embed_and_cluster.R

@@ -49,6 +49,26 @@ embed_and_cluster <-
     return(result)
   }
 
+#' Ensure is list in json
+#'
+#' When sending responses as json, Vectors of length 0 or 1 are converted to
+#' null and scalar (respectively) Using as.list fixes this, however, long R
+#' lists take a VERY long time to be converted to JSON.
+#' This function deals with the problematic cases, leaving vector as a vector
+#' when it isnt a problem.
+#'
+#' @param vector
+#'
+#' @export
+#'
+ensure_is_list_in_json <- function(vector) {
+  if (length(vector) <= 1) {
+    return(as.list(vector))
+  } else {
+    return(vector)
+  }
+}
+
 format_cell_sets_object <-
   function(cell_sets, clustering_method, color_pool) {
     name <- paste0(clustering_method, " clusters")
@@ -70,7 +90,7 @@ format_cell_sets_object <-
         rootNode = FALSE,
         type = "cellSets",
         color = color_pool[1],
-        cellIds = unname(cells)
+        cellIds = ensure_is_list_in_json(unname(cells))
       )
       color_pool <- color_pool[-1]
       cell_sets_object$children <-

pipeline-runner/R/subset-1-subset_seurat.R

@@ -1,3 +1,5 @@
+library(uuid)
+
 #' create a subset experiment
 #'
 #' This is the first step of a subset pipeline, which basically takes the parent