More granular control over NimbleAppend

Author: bbimber

singlecell/resources/chunks/AppendNimble.R

@@ -16,8 +16,8 @@ for (datasetId in names(seuratObjects)) {
   seuratObj <- readSeuratRDS(seuratObjects[[datasetId]])
 
   for (genomeId in names(nimbleGenomes)) {
-    retainAmbiguousFeatures <- !nimbleGenomeAmbiguousPreference[[genomeId]]
-    seuratObj <- Rdiscvr::DownloadAndAppendNimble(seuratObject = seuratObj, allowableGenomes = genomeId, ensureSamplesShareAllGenomes = ensureSamplesShareAllGenomes, targetAssayName = nimbleGenomes[[genomeId]], enforceUniqueFeatureNames = TRUE, dropAmbiguousFeatures = !retainAmbiguousFeatures, maxLibrarySizeRatio = maxLibrarySizeRatio)
+    maxAmbiguityAllowed <- !nimbleGenomeAmbiguousPreference[[genomeId]]
+    seuratObj <- Rdiscvr::DownloadAndAppendNimble(seuratObject = seuratObj, allowableGenomes = genomeId, ensureSamplesShareAllGenomes = ensureSamplesShareAllGenomes, targetAssayName = nimbleGenomes[[genomeId]], enforceUniqueFeatureNames = TRUE, maxAmbiguityAllowed = maxAmbiguityAllowed, maxLibrarySizeRatio = maxLibrarySizeRatio)
   }
 
   saveData(seuratObj, datasetId)

singlecell/resources/web/singlecell/panel/NimbleAppendPanel.js

@@ -40,7 +40,7 @@ Ext4.define('SingleCell.panel.NimbleAppendPanel', {
 				},LABKEY.ext4.GRIDBUTTONS.DELETERECORD()],
 				store: {
 					type: 'array',
-					fields: ['genomeId', 'targetAssay','retainAmbiguousFeatures']
+					fields: ['genomeId', 'targetAssay','maxAmbiguityAllowed']
 				},
 				columns: [{
 					dataIndex: 'genomeId',
@@ -69,12 +69,13 @@ Ext4.define('SingleCell.panel.NimbleAppendPanel', {
 						allowBlank: false
 					}
 				},{
-					dataIndex: 'retainAmbiguousFeatures',
+					dataIndex: 'maxAmbiguityAllowed',
 					width: 175,
-					header: 'Retain Ambiguous Features?',
+					header: 'Max Ambiguity Allowed',
 					editor: {
-						xtype: 'checkbox',
-						allowBlank: false
+						xtype: 'ldk-integerfield',
+						allowBlank: true,
+						minValue: 0
 					}
 				}]
 			}]
@@ -86,7 +87,7 @@ Ext4.define('SingleCell.panel.NimbleAppendPanel', {
 	getValue: function(){
 		var ret = [];
 		this.down('ldk-gridpanel').store.each(function(r, i) {
-			ret.push([r.data.genomeId, r.data.targetAssay, !!r.data.retainAmbiguousFeatures]);
+			ret.push([r.data.genomeId, r.data.targetAssay, r.data.maxAmbiguityAllowed ?? '']);
 		}, this);
 
 		return Ext4.isEmpty(ret) ? null : JSON.stringify(ret);
@@ -122,7 +123,7 @@ Ext4.define('SingleCell.panel.NimbleAppendPanel', {
 				var rec = grid.store.createModel({
 					genomeId: row[0],
 					targetAssay: row[1],
-					retainAmbiguousFeatures: !!row[2]
+					maxAmbiguityAllowed: row[2]
 				});
 				grid.store.add(rec);
 			}, this);

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

@@ -32,6 +32,10 @@ public Provider()
                     {{
                         put("allowBlank", false);
                     }}, null),
+                    SeuratToolParameter.create("maxAmbiguityAllowed", "Max Ambiguity Allowed", "If provided, ambiguous features with more than this number of values will be discarded (e.g. if maxAmbiguityAllowed=2, then the feature Feat1,Feat2,Feat3 would be discared, but not Feat1,Feat3. This can be overridden per genome.", "ldk-integerfield", new JSONObject()
+                    {{
+                        put("minValue", 0);
+                    }}, 0, null, true),
                     SeuratToolParameter.create("ensureSamplesShareAllGenomes", "Ensure Samples Share All Genomes", "If checked, the job will fail unless nimble data is found for each requested genome for all samples", "checkbox", new JSONObject()
                     {{
                         put("check", true);
@@ -84,6 +88,8 @@ protected Chunk createParamChunk(SequenceOutputHandler.JobContext ctx, List<Seur
         }
         ret.bodyLines.add(")");
 
+        Integer maxAmbiguityAllowed = getProvider().getParameterByName("maxAmbiguityAllowed").extractValue(getPipelineCtx().getJob(), getProvider(), getStepIdx(), Integer.class);
+
         ret.bodyLines.add("nimbleGenomeAmbiguousPreference <- list(");
         delim = "";
         for (int i = 0; i < json.length(); i++)
@@ -95,8 +101,13 @@ protected Chunk createParamChunk(SequenceOutputHandler.JobContext ctx, List<Seur
             }
 
             int genomeId = arr.getInt(0);
-            boolean retainAmbiguousFeatures = arr.getBoolean(2);
-            ret.bodyLines.add("\t" + delim + "'" + genomeId + "' = " + (retainAmbiguousFeatures ? "TRUE" : "FALSE"));
+            Integer maxAmbiguityAllowed2 = arr.get(2) == null ? null : arr.getInt(2);
+            if (maxAmbiguityAllowed2 == null)
+            {
+                maxAmbiguityAllowed2 = maxAmbiguityAllowed;
+            }
+
+            ret.bodyLines.add("\t" + delim + "'" + genomeId + "' = " + (maxAmbiguityAllowed2 == null ? "Inf" : maxAmbiguityAllowed2));
             delim = ",";
         }
         ret.bodyLines.add(")");