Add a TCR-specific seurat filter

Author: bbimber

singlecell/resources/chunks/TcrFilter.R

@@ -0,0 +1,48 @@
+totalPassed <- 0
+ for (datasetId in names(seuratObjects)) {
+    printName(datasetId)
+    seuratObj <- readRDS(seuratObjects[[datasetId]])
+
+    cellsToKeep <- NULL
+    for (locus in c('A', 'B', 'D', 'G')) {
+        fieldName <- paste0('TR', locus)
+
+        cdr3ForLocus <- cdr3s[grepl(cdr3s, pattern = paste0(fieldName, ':'))]
+        if (length(cdr3ForLocus) == 0) {
+            next
+        }
+
+        cdr3ForLocus <- gsub(cdr3ForLocus, pattern = paste0(fieldName, ':'), replacement = '')
+        matchingCells <- sapply(seuratObj@meta.data[[fieldName]], function(x){
+            values <- unlist(strsplit(x, split = ','))
+            return(length(intersect(values, cdr3ForLocus)) > 0)
+        })
+
+        if (sum(matchingCells) == 0) {
+            next
+        }
+
+        matchingCells <- colnames(seuratObj)[matchingCells]
+        if (all(is.null(cellsToKeep))) {
+            cellsToKeep <- matchingCells
+        } else {
+            cellsToKeep <- unique(c(cellsToKeep, matchingCells))
+        }
+    }
+    
+    if (all(is.null(cellsToKeep))) {
+        print('There were no matching cells')
+    } else {
+        seuratObj <- subset(seuratObj, cells = cellsToKeep)
+        saveData(seuratObj, datasetId)
+        totalPassed <- totalPassed + 1
+    }
+
+    # Cleanup
+    rm(seuratObj)
+    gc()
+}
+
+if (totalPassed == 0) {
+    addErrorMessage('No cells remained in any seurat objects after subsetting')
+}

singlecell/resources/web/singlecell/panel/NimbleAlignPanel.js

@@ -52,7 +52,7 @@ Ext4.define('SingleCell.panel.NimbleAlignPanel', {
 						const store = this.up('singlecell-nimblealignpanel').genomeField.store
 						if (val && store) {
 							const recIdx = store.find('rowid', val);
-							return store.getAt(recIdx).get('name');
+							return recIdx === -1 ? '[' + val + ']' : store.getAt(recIdx).get('name');
 						}
 						else if (val) {
 							return '[' + val + ']';

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

@@ -51,7 +51,7 @@ Ext4.define('SingleCell.panel.NimbleAppendPanel', {
 						const store = this.up('singlecell-nimbleappendpanel').genomeField.store
 						if (val && store) {
 							const recIdx = store.find('rowid', val);
-							return store.getAt(recIdx).get('name');
+							return recIdx === -1 ? '[' + val + ']' : store.getAt(recIdx).get('name');
 						}
 						else if (val) {
 							return '[' + val + ']';

singlecell/src/org/labkey/singlecell/SingleCellModule.java

@@ -190,6 +190,7 @@ public static void registerPipelineSteps()
         SequencePipelineService.get().registerPipelineStep(new RunVision.Provider());
         SequencePipelineService.get().registerPipelineStep(new NimbleAppend.Provider());
         SequencePipelineService.get().registerPipelineStep(new AppendTcr.Provider());
+        SequencePipelineService.get().registerPipelineStep(new TcrFilter.Provider());
 
         SequenceAnalysisService.get().registerFileHandler(new NimbleHandler());
     }

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

@@ -0,0 +1,44 @@
+package org.labkey.singlecell.pipeline.singlecell;
+
+import org.json.JSONObject;
+import org.labkey.api.sequenceanalysis.pipeline.AbstractPipelineStepProvider;
+import org.labkey.api.sequenceanalysis.pipeline.PipelineContext;
+import org.labkey.api.singlecell.pipeline.SeuratToolParameter;
+import org.labkey.api.singlecell.pipeline.SingleCellStep;
+
+import java.util.Arrays;
+
+public class TcrFilter extends AbstractCellMembraneStep
+{
+    public TcrFilter(PipelineContext ctx, TcrFilter.Provider provider)
+    {
+        super(provider, ctx);
+    }
+
+    public static class Provider extends AbstractPipelineStepProvider<SingleCellStep>
+    {
+        public Provider()
+        {
+            super("TcrFilter", "TCR-Based Filter", "CellMembrane/Seurat", "This will run standard Seurat processing steps to normalize and scale the data.", Arrays.asList(
+                    SeuratToolParameter.create("cdr3s", "CDR3s To Keep", "A list of CDR3 sequences where locus prefixes the AA sequence (i.e. TRB:XXXXXX or TRA:YYYYYYY). These should not contain commas. Any cell matching any of these CDR3s will be kept. If that cell has multiple chains for a locus (i.e. 'CASSXXXXX,CASSYYYYY'), then only one of these needs to match for that cell to be kept.", "sequenceanalysis-trimmingtextarea", new JSONObject(){{
+                        put("height", 150);
+                        put("delimiter", ",");
+                    }}, null).delimiter(",")
+            ), Arrays.asList("/sequenceanalysis/field/TrimmingTextArea.js"), null);
+        }
+
+        @Override
+        public TcrFilter create(PipelineContext ctx)
+        {
+            return new TcrFilter(ctx, this);
+        }
+    }
+
+    @Override
+    public String getFileSuffix()
+    {
+        return "tcrFilter";
+    }
+}
+
+