Clarify TCR filter

Author: bbimber

singlecell/resources/chunks/TcrFilter.R

@@ -1,48 +1,52 @@
 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))
-        }
+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
     }
-    
+
+    if (!(fieldName %in% names(seuratObj@meta.data))) {
+      stop(paste0('Missing field: ', fieldName))
+    }
+
+    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))) {
-        print('There were no matching cells')
+      cellsToKeep <- matchingCells
     } else {
-        seuratObj <- subset(seuratObj, cells = cellsToKeep)
-        saveData(seuratObj, datasetId)
-        totalPassed <- totalPassed + 1
+      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()
+  # Cleanup
+  rm(seuratObj)
+  gc()
 }
 
 if (totalPassed == 0) {
-    addErrorMessage('No cells remained in any seurat objects after subsetting')
+  addErrorMessage('No cells remained in any seurat objects after subsetting')
 }

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

@@ -20,7 +20,7 @@ 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(){{
+                    SeuratToolParameter.create("cdr3s", "CDR3s To Keep", "A comma- or newline-delimited 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(",")