Allow jobs to control more EmptyDrop options

Author: bbimber

singlecell/resources/chunks/PrepareRawCounts.R

@@ -4,6 +4,10 @@ for (datasetId in names(seuratObjects)) {
     datasetName <- datasetIdToName[[datasetId]]
     seuratObj <- CellMembrane::ReadAndFilter10xData(dataDir = rawCountDir, datasetId = datasetId, datasetName = datasetName, emptyDropsLower = emptyDropsLower, emptyDropsFdrThreshold = emptyDropsFdrThreshold)
 
+    if (!is.null(maxAllowableCells) && ncol(seuratObj) > maxAllowableCells) {
+        stop(paste0('The seurat object has more than the max allowable cells. Please review emptyDrops results'))
+    }
+
     saveData(seuratObj, datasetId)
 
     # Cleanup

singlecell/resources/web/singlecell/panel/SingleCellProcessingPanel.js

@@ -219,6 +219,13 @@ Ext4.define('SingleCell.panel.SingleCellProcessingPanel', {
                 defaultValue: 0.001,
                 minValue: 0,
                 decimalPrecision: 4
+            },{
+                fieldXtype: 'ldk-integerfield',
+                name: 'maxAllowableCells',
+                label: 'Max Cells Allowed',
+                description: 'If more than this many cells are predicted by EmptyDrops, the job will fail',
+                defaultValue: 20000,
+                minValue: 0
             }]
         }];
 

singlecell/src/org/labkey/singlecell/pipeline/singlecell/PrepareRawCounts.java

@@ -25,7 +25,8 @@ public Provider()
         {
             super("PrepareRawCounts", LABEL, "CellMembrane/Seurat", "This step reads the raw count matrix/matrices, and runs EmptyDrops to provide an unfiltered count matrix.", Arrays.asList(
                     SeuratToolParameter.create("emptyDropsLower", "Lower", "Passed to DropletUtils::emptyDrops lower argument", "ldk-integerfield", null, 200),
-                    SeuratToolParameter.create("emptyDropsFdrThreshold", "FDR Threshold", "The FDR limit used to filter the results of DropletUtils::emptyDrops", "ldk-numberfield", null, 0.001)
+                    SeuratToolParameter.create("emptyDropsFdrThreshold", "FDR Threshold", "The FDR limit used to filter the results of DropletUtils::emptyDrops", "ldk-numberfield", null, 0.001),
+                    SeuratToolParameter.create("maxAllowableCells", "Max Cells Allowed", "If more than this many cells are predicted by EmptyDrops, the job will fail", "ldk-integerfield", null, 20000)
             ), null, null);
         }