Small fix that manual outlier removal uses default assay.

Author: malteborggrewe

R/outlier_detection.R

@@ -18,11 +18,14 @@
 remove_outliers_manual <- function(seu,
                                    thresholds,
                                    negative = TRUE) {
+  # get expression matrix
+  data <- GetAssayData(seu)
+  # make list of cells to remove
   remove_cells <- c()
   for(marker in names(thresholds)) {
     # get cell names of cells not passing threshold
-    if(isTRUE(negative)) cells <- names(which(seu@assays$comp@data[marker,] < thresholds[[marker]]))
-    if(!isTRUE(negative)) cells <- names(which(seu@assays$comp@data[marker,] > thresholds[[marker]]))
+    if(isTRUE(negative)) cells <- names(which(data[marker,] < thresholds[[marker]]))
+    if(!isTRUE(negative)) cells <- names(which(data[marker,] > thresholds[[marker]]))
     # add cellnames to the array
     remove_cells <- c(remove_cells, cells)
   }
@@ -33,7 +36,6 @@ remove_outliers_manual <- function(seu,
   return(seu)
 }
 
-
 #' Labels potentially noisy or outlier events in Seurat object (seu$outlier)
 #' 
 #' Uses isolation forest to score events based on their probability to be an outlier. Scores and predicted outliers are stored in Seurat meta.data "outlier" and "outlier_score".