Add ability to conditionally use multiseq on large datasets

Author: bbimber

singlecell/api-src/org/labkey/api/singlecell/CellHashingService.java

@@ -153,7 +153,7 @@ public static CellHashingService.CellHashingParameters createFromStep(SequenceOu
                 if (methodStr2 != null)
                 {
                     ret.consensusMethods = extractMethods(methodStr2);
-                    if (!ret.methods.containsAll(ret.consensusMethods))
+                    if (!new HashSet<>(ret.methods).containsAll(ret.consensusMethods))
                     {
                         throw new PipelineJobException("All consensusMethods must be present in methods: " + methodStr2);
                     }
@@ -189,7 +189,7 @@ public static CellHashingParameters createFromJson(BARCODE_TYPE type, File webse
 
             if (ret.consensusMethods != null && !ret.consensusMethods.isEmpty())
             {
-                if (!ret.methods.containsAll(ret.consensusMethods))
+                if (!new HashSet<>(ret.methods).containsAll(ret.consensusMethods))
                 {
                     throw new PipelineJobException("All consensusMethods must be present in methods: " + ret.consensusMethods.stream().map(CALLING_METHOD::name).collect(Collectors.joining(",")));
                 }
@@ -326,6 +326,7 @@ public Set<String> getAllowableBarcodeNames() throws PipelineJobException
     public enum CALLING_METHOD
     {
         multiseq(true, false),
+        multiseqOnLargeData(true, true, false, 10000, "multiseq"),
         htodemux(false, false),
         dropletutils(true, true),
         gmm_demux(true, true),
@@ -337,17 +338,26 @@ public enum CALLING_METHOD
         boolean isDefaultRun;
         boolean isDefaultConsensus;
         boolean requiresH5;
+        int minCells;
+        String label;
 
         CALLING_METHOD(boolean isDefaultRun, boolean isDefaultConsensus)
         {
             this(isDefaultRun, isDefaultConsensus, false);
         }
 
         CALLING_METHOD(boolean isDefaultRun, boolean isDefaultConsensus, boolean requiresH5)
+        {
+            this(isDefaultRun, isDefaultConsensus, requiresH5, 0, null);
+        }
+
+        CALLING_METHOD(boolean isDefaultRun, boolean isDefaultConsensus, boolean requiresH5, int minCells, String label)
         {
             this.isDefaultRun = isDefaultRun;
             this.isDefaultConsensus = isDefaultConsensus;
             this.requiresH5 = requiresH5;
+            this.minCells = minCells;
+            this.label = label;
         }
 
         public boolean isDefaultRun()
@@ -360,6 +370,16 @@ public boolean isDefaultConsensus()
             return isDefaultConsensus;
         }
 
+        public int getMinCells()
+        {
+            return minCells;
+        }
+
+        public String getLabel()
+        {
+            return label == null ? name() : label;
+        }
+
         public boolean isRequiresH5()
         {
             return requiresH5;

singlecell/src/org/labkey/singlecell/CellHashingServiceImpl.java

@@ -69,6 +69,7 @@
 import java.util.concurrent.atomic.AtomicBoolean;
 import java.util.concurrent.atomic.AtomicInteger;
 import java.util.stream.Collectors;
+import java.util.stream.Stream;
 
 import static org.labkey.singlecell.run.CellRangerGexCountStep.LOUPE_CATEGORY;
 
@@ -1214,13 +1215,44 @@ public File generateCellHashingCalls(File citeSeqCountOutDir, File outputDir, St
         File localRScript = new File(outputDir, "generateCallsWrapper.R");
         try (PrintWriter writer = PrintWriters.getPrintWriter(localRScript))
         {
-            List<String> methodNames = parameters.methods.stream().map(Enum::name).collect(Collectors.toList());
-            List<String> consensusMethodNames = parameters.consensusMethods == null ? Collections.emptyList() : parameters.consensusMethods.stream().map(Enum::name).collect(Collectors.toList());
             String cellbarcodeWhitelist = cellBarcodeWhitelistFile != null ? "'" + cellBarcodeWhitelistFile.getPath() + "'" : "NULL";
+            long totalCellBarcodes;
+            if (cellBarcodeWhitelistFile != null)
+            {
+                try (Stream<String> st = Files.lines(cellBarcodeWhitelistFile.toPath()))
+                {
+                    totalCellBarcodes = st.count();
+                }
+            }
+            else
+            {
+                totalCellBarcodes = 99999L;
+            }
+            ctx.getLogger().debug("Total input cell barcodes: " + totalCellBarcodes);
 
             Set<String> allowableBarcodes = parameters.getAllowableBarcodeNames();
             String allowableBarcodeParam = allowableBarcodes != null ? "c('" + StringUtils.join(allowableBarcodes, "','") + "')" : "NULL";
 
+            List<String> methodNames = parameters.methods.stream().filter(m -> {
+                if (totalCellBarcodes > m.getMinCells())
+                {
+                    ctx.getLogger().debug("Dropping method due to insufficient cells: " + m.name());
+                    return false;
+                }
+
+                return true;
+            }).map(CALLING_METHOD::getLabel).distinct().toList();
+
+            List<String> consensusMethodNames = parameters.consensusMethods == null ? Collections.emptyList() : parameters.consensusMethods.stream().filter(m -> {
+                if (totalCellBarcodes > m.getMinCells())
+                {
+                    ctx.getLogger().debug("Dropping consensus method due to insufficient cells: " + m.name());
+                    return false;
+                }
+
+                return true;
+            }).map(CALLING_METHOD::getLabel).distinct().toList();
+
             String skipNormalizationQcString = parameters.skipNormalizationQc ? "TRUE" : "FALSE";
             String keepMarkdown = parameters.keepMarkdown ? "TRUE" : "FALSE";
             String doTSNE = parameters.doTSNE ? "TRUE" : "FALSE";