Allow many steps to operate on either BAM or CRAM

Author: bbimber

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

@@ -99,7 +99,7 @@ else if (sortOrder != SAMFileHeader.SortOrder.coordinate)
                 input.delete();
 
                 //note: if there is a pre-existing index, we need to delete this since it is out of date
-                File idx = new File(input.getPath() + ".bai");
+                File idx = SequencePipelineService.get().getExpectedIndex(output);
                 if (idx.exists())
                 {
                     getLogger().debug("deleting old BAM index");

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

@@ -22,9 +22,9 @@ public SamtoolsIndexer(Logger log)
 
     public File execute(File inputBam) throws PipelineJobException
     {
-        getLogger().info("Sorting SAM/BAM: " + inputBam.getPath());
+        getLogger().info("Indexing SAM/BAM: " + inputBam.getPath());
 
-        File idx = new File(inputBam.getPath() + ".bai");
+        File idx = SequencePipelineService.get().getExpectedIndex(inputBam);
         if (idx.exists())
         {
             getLogger().debug("deleting existing index: " + idx.getPath());

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

@@ -112,6 +112,8 @@ static public void setInstance(SequencePipelineService instance)
 
     abstract public long getLineCount(File f) throws PipelineJobException;
 
+    abstract public File getExpectedIndex(File bamOrCram);
+
     abstract public File ensureBamIndex(File f, Logger log, boolean forceDeleteExisting) throws PipelineJobException;
 
     abstract public SAMFileHeader.SortOrder getBamSortOrder(File bam) throws IOException;

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

@@ -61,7 +61,7 @@ public File execute(File inputFile, @Nullable File outputFile, SAMFileHeader.Sor
                 FileUtils.moveFile(outputBam, inputFile);
 
                 //note: if there is a pre-existing index, we need to delete this since it is out of date
-                File expectedIndex = new File(inputFile.getPath() + ".bai");
+                File expectedIndex = SequencePipelineService.get().getExpectedIndex(inputFile);
                 if (expectedIndex.exists())
                 {
                     getLogger().info("deleting out of date index: " + expectedIndex.getPath());

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

@@ -17,6 +17,7 @@
 import org.labkey.api.sequenceanalysis.pipeline.PipelineStepCtx;
 import org.labkey.api.sequenceanalysis.pipeline.PipelineStepProvider;
 import org.labkey.api.sequenceanalysis.pipeline.PreprocessingStep;
+import org.labkey.api.sequenceanalysis.pipeline.SamtoolsIndexer;
 import org.labkey.api.sequenceanalysis.pipeline.SequencePipelineService;
 import org.labkey.api.sequenceanalysis.pipeline.TaskFileManager;
 import org.labkey.api.sequenceanalysis.run.AbstractCommandWrapper;
@@ -437,10 +438,16 @@ public long getLineCount(File f) throws PipelineJobException
         return SequenceUtil.getLineCount(f);
     }
 
+    @Override
+    public File getExpectedIndex(File bamOrCram)
+    {
+        return SequenceUtil.getExpectedIndex(bamOrCram);
+    }
+
     @Override
     public File ensureBamIndex(File inputBam, Logger log, boolean forceDeleteExisting) throws PipelineJobException
     {
-        File expectedIndex = new File(inputBam.getPath() + ".bai");
+        File expectedIndex = SequenceUtil.getExpectedIndex(inputBam);
         if (expectedIndex.exists() && (expectedIndex.lastModified() < inputBam.lastModified() || forceDeleteExisting))
         {
             log.info("deleting out of date index: " + expectedIndex.getPath());
@@ -449,10 +456,8 @@ public File ensureBamIndex(File inputBam, Logger log, boolean forceDeleteExistin
 
         if (!expectedIndex.exists())
         {
-            log.debug("\tcreating temp index for BAM: " + inputBam.getName());
-            BuildBamIndexWrapper buildBamIndexWrapper = new BuildBamIndexWrapper(log);
-            buildBamIndexWrapper.setStringency(ValidationStringency.SILENT);
-            buildBamIndexWrapper.executeCommand(inputBam);
+            log.debug("\tcreating index for BAM: " + inputBam.getName());
+            new SamtoolsIndexer(log).execute(inputBam);
 
             return expectedIndex;
         }

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

@@ -3,8 +3,10 @@
 import org.apache.commons.io.FileUtils;
 import org.apache.logging.log4j.Logger;
 import org.labkey.api.pipeline.PipelineJobException;
+import org.labkey.api.sequenceanalysis.pipeline.SamtoolsIndexer;
 import org.labkey.api.sequenceanalysis.run.AbstractGatk4Wrapper;
 import org.labkey.sequenceanalysis.pipeline.ReblockGvcfHandler;
+import org.labkey.sequenceanalysis.util.SequenceUtil;
 
 import java.io.File;
 import java.io.IOException;
@@ -21,25 +23,22 @@ public HaplotypeCallerWrapper(Logger log)
         super(log);
     }
 
-    public void execute(File inputBam, File referenceFasta, File outputFile, List<String> options) throws PipelineJobException
+    public void execute(File inputBamOrCram, File referenceFasta, File outputFile, List<String> options) throws PipelineJobException
     {
-        execute(inputBam, referenceFasta, outputFile, options, true);
+        execute(inputBamOrCram, referenceFasta, outputFile, options, true);
     }
 
-    public void execute(File inputBam, File referenceFasta, File outputFile, List<String> options, boolean reblockGVCF) throws PipelineJobException
+    public void execute(File inputBamOrCram, File referenceFasta, File outputFile, List<String> options, boolean reblockGVCF) throws PipelineJobException
     {
-        getLogger().info("Running GATK 4 HaplotypeCaller for: " + inputBam.getName());
+        getLogger().info("Running GATK 4 HaplotypeCaller for: " + inputBamOrCram.getName());
 
         ensureDictionary(referenceFasta);
 
-        File expectedIndex = new File(inputBam.getPath() + ".bai");
-        boolean doDeleteIndex = false;
+        File expectedIndex = SequenceUtil.getExpectedIndex(inputBamOrCram);
         if (!expectedIndex.exists())
         {
-            getLogger().debug("\tcreating temp index for BAM: " + inputBam.getName());
-            new BuildBamIndexWrapper(getLogger()).executeCommand(inputBam);
-
-            doDeleteIndex = true;
+            getLogger().debug("\tcreating index for BAM: " + inputBamOrCram.getName());
+            new SamtoolsIndexer(getLogger()).execute(inputBamOrCram);
         }
         else
         {
@@ -51,7 +50,7 @@ public void execute(File inputBam, File referenceFasta, File outputFile, List<St
         args.add("-R");
         args.add(referenceFasta.getPath());
         args.add("-I");
-        args.add(inputBam.getPath());
+        args.add(inputBamOrCram.getPath());
         args.add("-O");
         args.add(outputFile.getPath());
         if (options != null)
@@ -71,12 +70,6 @@ public void execute(File inputBam, File referenceFasta, File outputFile, List<St
             throw new PipelineJobException("Expected output not found: " + outputFile.getPath());
         }
 
-        if (doDeleteIndex)
-        {
-            getLogger().debug("\tdeleting temp BAM index: " + expectedIndex.getPath());
-            expectedIndex.delete();
-        }
-
         if (reblockGVCF)
         {
             getLogger().info("Running GATK 4 ReblockGVCF for: " + outputFile.getName());

SequenceAnalysis/src/org/labkey/sequenceanalysis/util/SequenceUtil.java

@@ -82,6 +82,7 @@ public static enum FILETYPE
         fastq(Arrays.asList(".fastq", ".fq"), true),
         fasta(Arrays.asList(".fasta", ".fa", ".fna"), true),
         bam(".bam"),
+        cram(".cram"),
         sff(".sff"),
         gtf(Collections.singletonList(".gtf"), true),
         gff(Arrays.asList(".gff", ".gff3"), true),
@@ -394,36 +395,6 @@ public static JSONArray getReadGroupsForBam(File bam) throws IOException
         return ret;
     }
 
-    public static void deleteBamAndIndex(File bam)
-    {
-        bam.delete();
-
-        File idx = new File(bam.getPath() + ".bai");
-        if (idx.exists())
-        {
-            idx.delete();
-        }
-    }
-
-    public static void recreateOldBamIndex(File bam, boolean forceRecreate, @Nullable Logger log) throws PipelineJobException
-    {
-        File idx = new File(bam.getPath() + ".bai");
-
-        //delete out of date index
-        if (idx.exists() && (forceRecreate || idx.lastModified() < bam.lastModified()))
-        {
-            if (log != null)
-                log.info("deleting existing BAM index");
-
-            idx.delete();
-        }
-
-        if (!idx.exists())
-        {
-            new BuildBamIndexWrapper(log).executeCommand(bam);
-        }
-    }
-
     public static void sortROD(File input, Logger log, Integer startColumnIdx) throws IOException, PipelineJobException
     {
         boolean isCompressed = input.getPath().endsWith(".gz");
@@ -622,4 +593,18 @@ public static List<Interval> parseAndSortIntervals(String intervalString) throws
 
         return intervals;
     }
+
+    public static File getExpectedIndex(File bamOrCram)
+    {
+        if (FILETYPE.bam.getFileType().isType(bamOrCram))
+        {
+            return new File(bamOrCram.getPath() + ".bai");
+        }
+        else if (FILETYPE.cram.getFileType().isType(bamOrCram))
+        {
+            return new File(bamOrCram.getPath() + ".crai");
+        }
+
+        throw new IllegalArgumentException("Must provide either a .bam or .cram file");
+    }
 }