Merge pull request #338 from LabKey/fb_merge_24.11_to_develop

Merge discvr-24.11 to develop

Author: bbimber

SequenceAnalysis/api-src/org/labkey/api/sequenceanalysis/pipeline/BcftoolsRunner.java

@@ -2,9 +2,13 @@
 
 import org.apache.logging.log4j.Logger;
 import org.jetbrains.annotations.Nullable;
+import org.labkey.api.pipeline.PipelineJobException;
+import org.labkey.api.pipeline.PipelineJobService;
 import org.labkey.api.sequenceanalysis.run.AbstractCommandWrapper;
 
 import java.io.File;
+import java.util.ArrayList;
+import java.util.List;
 
 /**
  * User: bimber
@@ -22,4 +26,32 @@ public static File getBcfToolsPath()
     {
         return SequencePipelineService.get().getExeForPackage("BCFTOOLSPATH", "bcftools");
     }
+
+    public static boolean isBcftoolsFound()
+    {
+        return BcftoolsRunner.resolveFileInPath("bcftools", null, false) != null;
+    }
+
+    public void doIndex(File vcf) throws PipelineJobException
+    {
+        List<String> args = new ArrayList<>();
+        args.add(getBcfToolsPath().getAbsolutePath());
+        args.add("index");
+        args.add("-t");
+        args.add("-f");
+
+        if (!PipelineJobService.get().isWebServer())
+        {
+            Integer threads = SequencePipelineService.get().getMaxThreads(getLogger());
+            if (threads != null)
+            {
+                args.add("--threads");
+                args.add(String.valueOf(threads));
+            }
+        }
+
+        args.add(vcf.getAbsolutePath());
+
+        execute(args);
+    }
 }

SequenceAnalysis/src/org/labkey/sequenceanalysis/SequenceAnalysisServiceImpl.java

@@ -36,6 +36,7 @@
 import org.labkey.api.sequenceanalysis.SequenceAnalysisService;
 import org.labkey.api.sequenceanalysis.SequenceDataProvider;
 import org.labkey.api.sequenceanalysis.model.Readset;
+import org.labkey.api.sequenceanalysis.pipeline.BcftoolsRunner;
 import org.labkey.api.sequenceanalysis.pipeline.ReferenceGenome;
 import org.labkey.api.sequenceanalysis.pipeline.SamtoolsCramConverter;
 import org.labkey.api.sequenceanalysis.pipeline.SequenceOutputHandler;
@@ -267,8 +268,16 @@ public File ensureVcfIndex(File vcf, Logger log, boolean forceRecreate) throws I
                 //note: there is a bug in htsjdk's index creation with gz inputs
                 if (gz.isType(vcf) && !SystemUtils.IS_OS_WINDOWS)
                 {
-                    TabixRunner r = new TabixRunner(log);
-                    r.execute(vcf);
+                    // preferentially use bcftools since it supports multithreading:
+                    if (BcftoolsRunner.isBcftoolsFound())
+                    {
+                        new BcftoolsRunner(log).doIndex(vcf);
+                    }
+                    else
+                    {
+                        new TabixRunner(log).execute(vcf);
+                    }
+
                     if (!expectedIdx.exists())
                     {
                         throw new PipelineJobException("Expected index was not created: " + expectedIdx.getPath());

SequenceAnalysis/src/org/labkey/sequenceanalysis/analysis/LiftoverHandler.java

@@ -178,7 +178,7 @@ public void processFilesRemote(List<SequenceOutputFile> inputFiles, JobContext c
             boolean dropGenotypes = params.optBoolean("dropGenotypes", false);
             boolean useBcfTools = params.optBoolean("useBcfTools", false);
             boolean doNotRetainUnmapped = params.optBoolean("doNotRetainUnmapped", false);
-            if (doNotRetainUnmapped && !useBcfTools)
+            if (!doNotRetainUnmapped && !useBcfTools)
             {
                 ctx.getLogger().debug("Picard LiftoverVcf requires an output file for rejected sites, so setting doNotRetainUnmapped to true");
                 doNotRetainUnmapped = true;

SequenceAnalysis/src/org/labkey/sequenceanalysis/run/variant/GatherVcfsCloudWrapper.java

@@ -0,0 +1,54 @@
+package org.labkey.sequenceanalysis.run.variant;
+
+import org.apache.logging.log4j.Logger;
+import org.labkey.api.pipeline.PipelineJobException;
+import org.labkey.api.sequenceanalysis.SequenceAnalysisService;
+import org.labkey.api.sequenceanalysis.run.AbstractGatk4Wrapper;
+import org.labkey.api.writer.PrintWriters;
+
+import java.io.File;
+import java.io.IOException;
+import java.io.PrintWriter;
+import java.util.ArrayList;
+import java.util.List;
+
+public class GatherVcfsCloudWrapper extends AbstractGatk4Wrapper
+{
+    public GatherVcfsCloudWrapper(Logger log)
+    {
+        super(log);
+    }
+
+    public void gatherVcfs(File output, List<File> inputVcfs) throws PipelineJobException
+    {
+        List<String> args = new ArrayList<>(getBaseArgs("GatherVcfsCloud"));
+        args.add("-O");
+        args.add(output.getPath());
+
+        File argFile = new File(output.getParentFile(), "inputs.list");
+        try (PrintWriter writer = PrintWriters.getPrintWriter(argFile))
+        {
+            inputVcfs.forEach(f -> writer.println(f.getPath()));
+        }
+        catch (IOException e)
+        {
+            throw new PipelineJobException(e);
+        }
+
+        args.add("-I");
+        args.add(argFile.getPath());
+
+        execute(args);
+
+        argFile.delete();
+
+        try
+        {
+            SequenceAnalysisService.get().ensureVcfIndex(output, getLogger());
+        }
+        catch (IOException e)
+        {
+            throw new PipelineJobException(e);
+        }
+    }
+}