Merge pull request #345 from BimberLab/25.7_fb_merge

Merge discvr-25.3 to discvr-25.7

Author: bbimber

SequenceAnalysis/api-src/org/labkey/api/sequenceanalysis/run/AbstractGatk4Wrapper.java

@@ -91,6 +91,7 @@ public List<String> getBaseArgs(@Nullable String toolName)
         args.add(SequencePipelineService.get().getJavaFilepath());
         args.addAll(SequencePipelineService.get().getJavaOpts(_maxRamOverride));
         args.add("-DGATK_STACKTRACE_ON_USER_EXCEPTION=true");
+        args.add("-Dsamjdk.optimistic_vcf_4_4=true");
         args.add("-jar");
         args.add(getJAR().getPath());
 

SequenceAnalysis/build.gradle

@@ -135,7 +135,7 @@ dependencies {
     BuildUtils.addExternalDependency(
             project,
             new ExternalDependency(
-                    "com.github.broadinstitute:picard:3.1.0",
+                    "com.github.broadinstitute:picard:3.4.0",
                     "Picard Tools Lib",
                     "PicardTools",
                     "https://github.com/broadinstitute/picard",

SequenceAnalysis/src/org/labkey/sequenceanalysis/run/util/FastqcRunner.java

@@ -17,16 +17,16 @@
 
 import org.apache.commons.io.IOUtils;
 import org.apache.commons.lang3.StringUtils;
-import org.apache.logging.log4j.Logger;
 import org.apache.logging.log4j.LogManager;
+import org.apache.logging.log4j.Logger;
 import org.jetbrains.annotations.Nullable;
 import org.junit.Assert;
 import org.junit.Test;
 import org.labkey.api.module.Module;
 import org.labkey.api.module.ModuleLoader;
 import org.labkey.api.pipeline.PipelineJobService;
-import org.labkey.api.resource.FileResource;
 import org.labkey.api.resource.DirectoryResource;
+import org.labkey.api.resource.FileResource;
 import org.labkey.api.resource.Resource;
 import org.labkey.api.sequenceanalysis.pipeline.SequencePipelineService;
 import org.labkey.api.settings.AppProps;
@@ -46,6 +46,7 @@
 import java.io.InputStreamReader;
 import java.io.StringWriter;
 import java.util.ArrayList;
+import java.util.Arrays;
 import java.util.HashSet;
 import java.util.LinkedList;
 import java.util.List;
@@ -366,7 +367,7 @@ private List<String> getBaseParams() throws FileNotFoundException
             throw new RuntimeException("Not found: " + jbzip2.getPath());
         }
 
-        File htsjdkJar = new File(libDir, "htsjdk-4.0.0.jar");
+        File htsjdkJar = findJar(libDir, "htsjdk-");
         if (!htsjdkJar.exists())
         {
             throw new RuntimeException("Not found: " + htsjdkJar.getPath());
@@ -401,6 +402,27 @@ private List<String> getBaseParams() throws FileNotFoundException
         return params;
     }
 
+    private File findJar(final File libDir, final String prefix)
+    {
+        if (!libDir.exists())
+        {
+            throw new RuntimeException("Missing directory: " + libDir);
+        }
+
+        List<String> jarNames = Arrays.stream(libDir.list()).filter(fn -> fn.startsWith(prefix)).sorted().toList();
+        if (jarNames.isEmpty())
+        {
+            throw new RuntimeException("Unable to find JAR with prefix: " + prefix);
+        }
+
+        if (jarNames.size() > 1)
+        {
+            _logger.info("More than one JAR found with prefix: " + prefix);
+        }
+
+        return new File(libDir, jarNames.get(jarNames.size() - 1));
+    }
+
     private int getThreads()
     {
         return _threads;

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

@@ -115,6 +115,10 @@ public Output processVariants(File inputVCF, File outputDirectory, ReferenceGeno
         plinkArgs.add("--max-alleles");
         plinkArgs.add("2");
 
+        // NOTE: tools like sawfish can report half-called genotypes, like 0/.. For now, be most conservative in PCA:
+        plinkArgs.add("--vcf-half-call");
+        plinkArgs.add("missing");
+
         Integer threads = SequencePipelineService.get().getMaxThreads(getPipelineCtx().getLogger());
         if (threads != null)
         {

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

@@ -227,6 +227,10 @@ private void runBatch(File inputVCF, File outputDirectory, VariantProcessingStep
             args.add(String.valueOf(maxRam));
         }
 
+        // NOTE: tools like sawfish can report half-called genotypes, like 0/.. For now, be most conservative in PCA:
+        args.add("--vcf-half-call");
+        args.add("missing");
+
         args.addAll(getClientCommandArgs());
 
         getWrapper().execute(args);

singlecell/resources/chunks/ApplyKnownClontypicData.R

@@ -0,0 +1,20 @@
+netRc <- paste0(Sys.getenv('USER_HOME'), '/.netrc')
+if (!file.exists(netRc)) {
+    print(list.files(Sys.getenv('USER_HOME')))
+    stop(paste0('Unable to find file: ', netRc))
+}
+
+invisible(Rlabkey::labkey.setCurlOptions(NETRC_FILE = netRc, connect-timeout = 10))
+Rdiscvr::SetLabKeyDefaults(baseUrl = serverBaseUrl, defaultFolder = defaultLabKeyFolder)
+
+for (datasetId in names(seuratObjects)) {
+    printName(datasetId)
+    seuratObj <- readSeuratRDS(seuratObjects[[datasetId]])
+
+    seuratObj <- ApplyKnownClontypicData(seuratObj, groupFields = groupFields, addMetadata = addMetadata)
+    saveData(seuratObj, datasetId)
+
+    # Cleanup
+    rm(seuratObj)
+    gc()
+}

singlecell/src/org/labkey/singlecell/SingleCellModule.java

@@ -42,6 +42,7 @@
 import org.labkey.singlecell.pipeline.singlecell.AppendNimble;
 import org.labkey.singlecell.pipeline.singlecell.AppendSaturation;
 import org.labkey.singlecell.pipeline.singlecell.AppendTcr;
+import org.labkey.singlecell.pipeline.singlecell.ApplyKnownClonotypicData;
 import org.labkey.singlecell.pipeline.singlecell.AvgExpression;
 import org.labkey.singlecell.pipeline.singlecell.CalculateGeneComponentScores;
 import org.labkey.singlecell.pipeline.singlecell.CalculateUCellScores;
@@ -299,6 +300,7 @@ public static void registerPipelineSteps()
         SequencePipelineService.get().registerPipelineStep(new PerformDefaultNimbleAppend.Provider());
         SequencePipelineService.get().registerPipelineStep(new PerformMhcDimRedux.Provider());
         SequencePipelineService.get().registerPipelineStep(new RunTricycle.Provider());
+        SequencePipelineService.get().registerPipelineStep(new ApplyKnownClonotypicData.Provider());
 
         SequenceAnalysisService.get().registerReadsetListener(new SingleCellReadsetListener());
     }

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

@@ -0,0 +1,39 @@
+package org.labkey.singlecell.pipeline.singlecell;
+
+import org.labkey.api.sequenceanalysis.pipeline.AbstractPipelineStepProvider;
+import org.labkey.api.sequenceanalysis.pipeline.PipelineContext;
+import org.labkey.api.singlecell.pipeline.SingleCellStep;
+
+import java.util.List;
+
+public class ApplyKnownClonotypicData extends AbstractRDiscvrStep
+{
+    public ApplyKnownClonotypicData(PipelineContext ctx, ApplyKnownClonotypicData.Provider provider)
+    {
+        super(provider, ctx);
+    }
+
+    public static class Provider extends AbstractPipelineStepProvider<SingleCellStep>
+    {
+        public Provider()
+        {
+            super("ApplyKnownClonotypicData", "Append Known Clonotype/Antigen Data", "RDiscvr", "This will query the clone_responses table and append a column tagging each cell for matching antigens (based on clonotype)", List.of(
+
+            ), null, null);
+        }
+
+
+        @Override
+        public ApplyKnownClonotypicData create(PipelineContext ctx)
+        {
+            return new ApplyKnownClonotypicData(ctx, this);
+        }
+    }
+
+    @Override
+    public String getFileSuffix()
+    {
+        return "ctd";
+    }
+}
+