Merge 25.3 to develop

SequenceAnalysis/api-src/org/labkey/api/sequenceanalysis/run/AbstractCommandWrapper.java

@@ -44,6 +44,7 @@ abstract public class AbstractCommandWrapper implements CommandWrapper
     private File _outputDir = null;
     private File _workingDir = null;
     private Logger _log;
+    private boolean _logPath = false;
     private Level _logLevel = Level.DEBUG;
     private boolean _warnNonZeroExits = true;
     private boolean _throwNonZeroExits = true;
@@ -205,9 +206,11 @@ private void setPath(ProcessBuilder pb)
         {
             String path = System.getenv("PATH");
 
-            getLogger().debug("Existing PATH: " + path);
-            getLogger().debug("toolDir: " + toolDir);
-
+            if (_logPath)
+            {
+                getLogger().debug("Existing PATH: " + path);
+                getLogger().debug("toolDir: " + toolDir);
+            }
 
             if (path == null)
             {
@@ -229,11 +232,19 @@ private void setPath(ProcessBuilder pb)
                     path = fileExe.getParent() + File.pathSeparatorChar + path;
             }
 
-            getLogger().debug("using path: " + path);
+            if (_logPath)
+            {
+                getLogger().debug("using path: " + path);
+            }
             pb.environment().put("PATH", path);
         }
     }
 
+    public void setLogPath(boolean logPath)
+    {
+        _logPath = logPath;
+    }
+
     public void setOutputDir(File outputDir)
     {
         _outputDir = outputDir;

SequenceAnalysis/pipeline_code/extra_tools_install.sh

@@ -319,3 +319,17 @@ then
 else
     echo "Already installed"
 fi
+
+if [[ ! -e ${LKTOOLS_DIR}/sawfish || ! -z $FORCE_REINSTALL ]];
+then
+    echo "Cleaning up previous installs"
+    rm -Rf $LKTOOLS_DIR/sawfish*
+
+    wget https://github.com/PacificBiosciences/sawfish/releases/download/v2.0.0/sawfish-v2.0.0-x86_64-unknown-linux-gnu.tar.gz
+    tar -xzf sawfish-v2.0.0-x86_64-unknown-linux-gnu.tar.gz
+
+    mv sawfish-v2.0.0-x86_64-unknown-linux-gnu $LKTOOLS_DIR/
+    ln -s $LKTOOLS_DIR/sawfish-v2.0.0/bin/sawfish $LKTOOLS_DIR/
+else
+    echo "Already installed"
+fi

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

@@ -123,6 +123,8 @@
 import org.labkey.sequenceanalysis.run.analysis.PbsvAnalysis;
 import org.labkey.sequenceanalysis.run.analysis.PbsvJointCallingHandler;
 import org.labkey.sequenceanalysis.run.analysis.PindelAnalysis;
+import org.labkey.sequenceanalysis.run.analysis.SawfishAnalysis;
+import org.labkey.sequenceanalysis.run.analysis.SawfishJointCallingHandler;
 import org.labkey.sequenceanalysis.run.analysis.SequenceBasedTypingAnalysis;
 import org.labkey.sequenceanalysis.run.analysis.SnpCountAnalysis;
 import org.labkey.sequenceanalysis.run.analysis.SubreadAnalysis;
@@ -342,6 +344,7 @@ public static void registerPipelineSteps()
         SequencePipelineService.get().registerPipelineStep(new PindelAnalysis.Provider());
         SequencePipelineService.get().registerPipelineStep(new PbsvAnalysis.Provider());
         SequencePipelineService.get().registerPipelineStep(new GenrichStep.Provider());
+        SequencePipelineService.get().registerPipelineStep(new SawfishAnalysis.Provider());
 
         SequencePipelineService.get().registerPipelineStep(new PARalyzerAnalysis.Provider());
         SequencePipelineService.get().registerPipelineStep(new RnaSeQCStep.Provider());
@@ -400,6 +403,7 @@ public static void registerPipelineSteps()
         SequenceAnalysisService.get().registerFileHandler(new NextCladeHandler());
         SequenceAnalysisService.get().registerFileHandler(new ConvertToCramHandler());
         SequenceAnalysisService.get().registerFileHandler(new PbsvJointCallingHandler());
+        SequenceAnalysisService.get().registerFileHandler(new SawfishJointCallingHandler());
         SequenceAnalysisService.get().registerFileHandler(new DeepVariantHandler());
         SequenceAnalysisService.get().registerFileHandler(new GLNexusHandler());
         SequenceAnalysisService.get().registerFileHandler(new ParagraphStep());

SequenceAnalysis/src/org/labkey/sequenceanalysis/run/analysis/SawfishAnalysis.java

@@ -0,0 +1,163 @@
+package org.labkey.sequenceanalysis.run.analysis;
+
+import org.apache.logging.log4j.Logger;
+import org.jetbrains.annotations.Nullable;
+import org.labkey.api.pipeline.PipelineJobException;
+import org.labkey.api.sequenceanalysis.model.AnalysisModel;
+import org.labkey.api.sequenceanalysis.model.Readset;
+import org.labkey.api.sequenceanalysis.pipeline.AbstractAnalysisStepProvider;
+import org.labkey.api.sequenceanalysis.pipeline.AbstractPipelineStep;
+import org.labkey.api.sequenceanalysis.pipeline.AnalysisOutputImpl;
+import org.labkey.api.sequenceanalysis.pipeline.AnalysisStep;
+import org.labkey.api.sequenceanalysis.pipeline.PipelineContext;
+import org.labkey.api.sequenceanalysis.pipeline.PipelineStepProvider;
+import org.labkey.api.sequenceanalysis.pipeline.ReferenceGenome;
+import org.labkey.api.sequenceanalysis.pipeline.SamtoolsIndexer;
+import org.labkey.api.sequenceanalysis.pipeline.SamtoolsRunner;
+import org.labkey.api.sequenceanalysis.pipeline.SequencePipelineService;
+import org.labkey.api.sequenceanalysis.run.SimpleScriptWrapper;
+import org.labkey.sequenceanalysis.util.SequenceUtil;
+
+import java.io.File;
+import java.util.ArrayList;
+import java.util.List;
+
+public class SawfishAnalysis extends AbstractPipelineStep implements AnalysisStep
+{
+    public SawfishAnalysis(PipelineStepProvider<?> provider, PipelineContext ctx)
+    {
+        super(provider, ctx);
+    }
+
+    public static class Provider extends AbstractAnalysisStepProvider<SawfishAnalysis>
+    {
+        public Provider()
+        {
+            super("sawfish", "Sawfish Analysis", null, "This will run sawfish SV dicvoery and calling on the selected BAMs", List.of(), null, null);
+        }
+
+
+        @Override
+        public SawfishAnalysis create(PipelineContext ctx)
+        {
+            return new SawfishAnalysis(this, ctx);
+        }
+    }
+
+    @Override
+    public Output performAnalysisPerSampleRemote(Readset rs, File inputBam, ReferenceGenome referenceGenome, File outputDir) throws PipelineJobException
+    {
+        AnalysisOutputImpl output = new AnalysisOutputImpl();
+
+        File inputFile = inputBam;
+        if (SequenceUtil.FILETYPE.cram.getFileType().isType(inputFile))
+        {
+            CramToBam samtoolsRunner = new CramToBam(getPipelineCtx().getLogger());
+            File bam = new File(getPipelineCtx().getWorkingDirectory(), inputFile.getName().replaceAll(".cram$", ".bam"));
+            File bamIdx = new File(bam.getPath() + ".bai");
+            if (!bamIdx.exists())
+            {
+                samtoolsRunner.convert(inputFile, bam, referenceGenome.getWorkingFastaFile(), SequencePipelineService.get().getMaxThreads(getPipelineCtx().getLogger()));
+                new SamtoolsIndexer(getPipelineCtx().getLogger()).execute(bam);
+            }
+            else
+            {
+                getPipelineCtx().getLogger().debug("BAM index exists, will not re-convert CRAM");
+            }
+
+            inputFile = bam;
+
+            output.addIntermediateFile(bam);
+            output.addIntermediateFile(bamIdx);
+        }
+
+        List<String> args = new ArrayList<>();
+        args.add(getExe().getPath());
+        args.add("discover");
+
+        args.add("--bam");
+        args.add(inputFile.getPath());
+
+        args.add("--ref");
+        args.add(referenceGenome.getWorkingFastaFile().getPath());
+
+        File svOutDir = new File(outputDir, "sawfish");
+        args.add("--output-dir");
+        args.add(svOutDir.getPath());
+
+        Integer maxThreads = SequencePipelineService.get().getMaxThreads(getPipelineCtx().getLogger());
+        if (maxThreads != null)
+        {
+            args.add("--threads");
+            args.add(String.valueOf(maxThreads));
+        }
+
+        File bcf = new File(svOutDir, "candidate.sv.bcf");
+        File bcfIdx = new File(bcf.getPath() + ".csi");
+        if (bcfIdx.exists())
+        {
+            getPipelineCtx().getLogger().debug("BCF index already exists, reusing output");
+        }
+        else
+        {
+            new SimpleScriptWrapper(getPipelineCtx().getLogger()).execute(args);
+        }
+
+        if (!bcf.exists())
+        {
+            throw new PipelineJobException("Unable to find file: " + bcf.getPath());
+        }
+
+        output.addSequenceOutput(bcf, rs.getName() + ": sawfish", "Sawfish SV Discovery", rs.getReadsetId(), null, referenceGenome.getGenomeId(), null);
+
+        return output;
+    }
+
+    @Override
+    public Output performAnalysisPerSampleLocal(AnalysisModel model, File inputBam, File referenceFasta, File outDir) throws PipelineJobException
+    {
+        return null;
+    }
+
+    private File getExe()
+    {
+        return SequencePipelineService.get().getExeForPackage("SAWFISHPATH", "sawfish");
+    }
+
+    private static class CramToBam extends SamtoolsRunner
+    {
+        public CramToBam(Logger log)
+        {
+            super(log);
+        }
+
+        public void convert(File inputCram, File outputBam, File fasta, @Nullable Integer threads) throws PipelineJobException
+        {
+            getLogger().info("Converting CRAM to BAM");
+
+            execute(getParams(inputCram, outputBam, fasta, threads));
+        }
+
+        private List<String> getParams(File inputCram, File outputBam, File fasta, @Nullable Integer threads)
+        {
+            List<String> params = new ArrayList<>();
+            params.add(getSamtoolsPath().getPath());
+            params.add("view");
+            params.add("-b");
+            params.add("-T");
+            params.add(fasta.getPath());
+            params.add("-o");
+            params.add(outputBam.getPath());
+
+            if (threads != null)
+            {
+                params.add("-@");
+                params.add(String.valueOf(threads));
+            }
+
+            params.add(inputCram.getPath());
+
+            return params;
+        }
+    }
+}