Bugfix to handling of 10x barcodes

Author: bbimber

singlecell/src/org/labkey/singlecell/run/CellRangerGexCountStep.java

@@ -400,6 +400,10 @@ public AlignmentOutput performAlignment(Readset rs, List<File> inputFastqs1, @Nu
         {
             NimbleHelper.write10xBarcodes(bam, getWrapper().getLogger(), rs, referenceGenome, output);
         }
+        else
+        {
+            getPipelineCtx().getLogger().info("BAM file does not exist, cannot write 10x barcodes: " + bam.getPath());
+        }
 
         if (!shouldDiscardBam())
         {

singlecell/src/org/labkey/singlecell/run/NimbleAlignmentStep.java

@@ -41,6 +41,7 @@ public class NimbleAlignmentStep extends AbstractCellRangerDependentStep
     public static final String MAX_HITS_TO_REPORT = "maxHitsToReport";
     public static final String STRANDEDNESS = "strandedness";
     public static final String REQUIRE_CACHED_BARCODES = "requireCachedBarcodes";
+    public static final String WRITE_10X_BARCODES = "write10xBarcodes";
 
     public NimbleAlignmentStep(AlignmentStepProvider<?> provider, PipelineContext ctx, CellRangerWrapper wrapper)
     {
@@ -76,7 +77,10 @@ public static List<ToolParameterDescriptor> getToolParameters()
                 }}, 4),
                 ToolParameterDescriptor.create(REQUIRE_CACHED_BARCODES, "Fail Unless Cached Barcodes Present", "If checked, the pipeline will expect a previously computed map of cellbarcodes and UMIs to be computed. Under default conditions, if this is missing, cellranger will be re-run. This flag can be helpful to avoid that computation if you expect the barcode file to exist.", "checkbox", new JSONObject(){{
 
-                }}, false)
+                }}, false),
+                ToolParameterDescriptor.create(WRITE_10X_BARCODES, "Write 10x Barcodes To File", "If checked, the pipeline will save the 10x barcodes from the BAM to a TSV. This facilitates future analyses.", "checkbox", new JSONObject(){{
+
+                }}, true)
         );
     }
 

singlecell/src/org/labkey/singlecell/run/NimbleAnalysis.java

@@ -1,8 +1,5 @@
 package org.labkey.singlecell.run;
 
-import htsjdk.samtools.SAMFileHeader;
-import htsjdk.samtools.SAMTag;
-import htsjdk.samtools.SamReaderFactory;
 import org.labkey.api.pipeline.PipelineJobException;
 import org.labkey.api.sequenceanalysis.model.AnalysisModel;
 import org.labkey.api.sequenceanalysis.model.Readset;
@@ -21,6 +18,8 @@
 import java.util.LinkedHashSet;
 import java.util.List;
 
+import static org.labkey.singlecell.run.NimbleAlignmentStep.WRITE_10X_BARCODES;
+
 public class NimbleAnalysis extends AbstractPipelineStep implements AnalysisStep
 {
     public NimbleAnalysis(PipelineStepProvider<?> provider, PipelineContext ctx)
@@ -61,14 +60,14 @@ public Output performAnalysisPerSampleRemote(Readset rs, File inputBam, Referenc
         NimbleHelper helper = new NimbleHelper(getPipelineCtx(), getProvider(), getStepIdx());
         helper.doNimbleAlign(inputBam, output, rs, FileUtil.getBaseName(inputBam));
 
-        SAMFileHeader header = SamReaderFactory.makeDefault().getFileHeader(inputBam);
-        if (header.getAttribute(SAMTag.CB.name()) != null)
+        boolean write10xBarcodes = getProvider().getParameterByName(WRITE_10X_BARCODES).extractValue(getPipelineCtx().getJob(), getProvider(), getStepIdx(), Boolean.class, true);
+        if (write10xBarcodes)
         {
             NimbleHelper.write10xBarcodes(inputBam, getPipelineCtx().getLogger(), rs, referenceGenome, output);
         }
         else
         {
-            getPipelineCtx().getLogger().info("BAM lacks CB tag, will not output 10x barcodes to file");
+            getPipelineCtx().getLogger().info("10x barcodes will not be saved to TSV");
         }
 
         return output;