Improve SnpEff check for existing index

Author: bbimber

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

@@ -15,6 +15,7 @@
 import org.labkey.api.sequenceanalysis.pipeline.VariantProcessingStep;
 import org.labkey.api.sequenceanalysis.pipeline.VariantProcessingStepOutputImpl;
 import org.labkey.api.sequenceanalysis.run.AbstractCommandPipelineStep;
+import org.labkey.api.util.FileUtil;
 import org.labkey.api.util.PageFlowUtil;
 import org.labkey.api.writer.PrintWriters;
 import org.labkey.sequenceanalysis.pipeline.SequenceTaskHelper;
@@ -75,18 +76,16 @@ public static File checkOrCreateIndex(SequenceAnalysisJobSupport support, Logger
 
         SnpEffWrapper wrapper = new SnpEffWrapper(log);
         File snpEffIndexDir = wrapper.getExpectedIndexDir(snpEffBaseDir, genome.getGenomeId(), geneFileId);
-        if (snpEffIndexDir.exists())
-        {
-            log.debug("previously created index found, re-using: " + snpEffIndexDir.getPath());
-            return snpEffBaseDir;
-        }
-
-        File binFile = new File(snpEffIndexDir, "snpEffectPredictor.bin");
+        File binFile = FileUtil.appendName(snpEffIndexDir, "snpEffectPredictor.bin");
         if (!binFile.exists())
         {
             log.debug("existing index not found, expected: " + binFile.getPath());
             wrapper.buildIndex(snpEffBaseDir, genome, geneFile, geneFileId);
         }
+        else
+        {
+            log.debug("previously created index found, re-using: " + snpEffIndexDir.getPath());
+        }
 
         return snpEffBaseDir;
     }
@@ -100,7 +99,7 @@ public Output processVariants(File inputVCF, File outputDirectory, ReferenceGeno
         Integer geneFileId = getProvider().getParameterByName(GENE_PARAM).extractValue(getPipelineCtx().getJob(), getProvider(), getStepIdx(), Integer.class);
         File snpEffBaseDir = checkOrCreateIndex(getPipelineCtx().getSequenceSupport(), getPipelineCtx().getLogger(), genome, geneFileId);
 
-        File outputVcf = new File(outputDirectory, SequenceTaskHelper.getUnzippedBaseName(inputVCF) + ".snpEff.vcf.gz");
+        File outputVcf = FileUtil.appendName(outputDirectory, SequenceTaskHelper.getUnzippedBaseName(inputVCF) + ".snpEff.vcf.gz");
         if (outputVcf.exists())
         {
             getPipelineCtx().getLogger().debug("deleting pre-existing output file: " + outputVcf.getPath());
@@ -110,7 +109,7 @@ public Output processVariants(File inputVCF, File outputDirectory, ReferenceGeno
         File intFile = null;
         if (intervals != null)
         {
-            intFile = new File(outputVcf.getParentFile(), "snpEffintervals.bed");
+            intFile = FileUtil.appendName(outputVcf.getParentFile(), "snpEffintervals.bed");
             try (PrintWriter writer = PrintWriters.getPrintWriter(intFile))
             {
                 getPipelineCtx().getLogger().debug("Adding SnpEff intervals: " + intervals.size());

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

@@ -94,15 +94,16 @@ public String getGenomeBasename(Integer genomeId, Integer geneFileId)
     public File getExpectedIndexDir(File snpEffBaseDir, Integer genomeId, Integer geneFileId)
     {
         String basename = getGenomeBasename(genomeId, geneFileId);
-        return new File(snpEffBaseDir, basename);
+        return FileUtil.appendName(snpEffBaseDir, basename);
     }
 
     public void buildIndex(File snpEffBaseDir, ReferenceGenome genome, File genes, Integer geneFileId) throws PipelineJobException
     {
         getLogger().info("Building SnpEff index for: "+ genome.getGenomeId() + " / " + geneFileId);
 
         File genomeDir = getExpectedIndexDir(snpEffBaseDir, genome.getGenomeId(), geneFileId);
-        if (genomeDir.exists() && genomeDir.list().length > 0)
+        File doneFile = FileUtil.appendName(snpEffBaseDir, "build.done");
+        if (doneFile.exists())
         {
             getLogger().info("directory already exists, will not re-build");
             return;
@@ -144,8 +145,8 @@ else if ("gbk".equalsIgnoreCase(ext))
 
         try
         {
-            Files.createSymbolicLink(new File(genomeDir, "sequences.fa").toPath(), genome.getSourceFastaFile().toPath());
-            Files.createSymbolicLink(new File(genomeDir, "genes." + ext).toPath(), genes.toPath());
+            Files.createSymbolicLink(FileUtil.appendName(genomeDir, "sequences.fa").toPath(), genome.getSourceFastaFile().toPath());
+            Files.createSymbolicLink(FileUtil.appendName(genomeDir, "genes." + ext).toPath(), genes.toPath());
         }
         catch (IOException e)
         {
@@ -183,11 +184,11 @@ private File getJarDir()
 
     private File getSnpEffJar()
     {
-        return new File(getJarDir(), "snpEff.jar");
+        return FileUtil.appendName(getJarDir(), "snpEff.jar");
     }
 
     private File getSnpEffConfigFile()
     {
-        return new File(getJarDir(), "snpEff.config");
+        return FileUtil.appendName(getJarDir(), "snpEff.config");
     }
 }