From 02e23e2d9cf9c6496f5681b39a288334cf3dbb47 Mon Sep 17 00:00:00 2001
From: hanna <hanna@348d0f76-0448-11de-a6fe-93d51630548a>
Date: Wed, 13 Jan 2010 02:42:16 +0000
Subject: [PATCH] Threading support for beagle output files.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2569 348d0f76-0448-11de-a6fe-93d51630548a
---
 .../genotyper/GenotypeCalculationModel.java   | 23 ++-------
 .../GenotypeCalculationModelFactory.java      |  6 +--
 ...JointEstimateGenotypeCalculationModel.java | 10 ----
 .../walkers/genotyper/UnifiedGenotyper.java   | 49 +++++++++----------
 4 files changed, 29 insertions(+), 59 deletions(-)

diff --git a/java/src/org/broadinstitute/sting/gatk/walkers/genotyper/GenotypeCalculationModel.java b/java/src/org/broadinstitute/sting/gatk/walkers/genotyper/GenotypeCalculationModel.java
index c8173b03b..d85bbd6bc 100755
--- a/java/src/org/broadinstitute/sting/gatk/walkers/genotyper/GenotypeCalculationModel.java
+++ b/java/src/org/broadinstitute/sting/gatk/walkers/genotyper/GenotypeCalculationModel.java
@@ -35,8 +35,8 @@ public abstract class GenotypeCalculationModel implements Cloneable {
     protected double CONFIDENCE_THRESHOLD;
     protected double MINIMUM_ALLELE_FREQUENCY;
     protected boolean REPORT_SLOD;
-    protected PrintWriter verboseWriter;
-    protected PrintWriter beagleWriter;
+    protected PrintStream verboseWriter;
+    protected PrintStream beagleWriter;
 
     /**
      * Create a new GenotypeCalculationModel object
@@ -58,8 +58,8 @@ public abstract class GenotypeCalculationModel implements Cloneable {
                               Logger logger,
                               UnifiedArgumentCollection UAC,
                               GenotypeWriterFactory.GENOTYPE_FORMAT outputFormat,
-                              PrintWriter verboseWriter,
-                              PrintWriter beagleWriter) {
+                              PrintStream verboseWriter,
+                              PrintStream beagleWriter) {
         this.samples = new TreeSet<String>(samples);
         this.logger = logger;
         baseModel = UAC.baseModel;
@@ -73,22 +73,7 @@ public abstract class GenotypeCalculationModel implements Cloneable {
         MINIMUM_ALLELE_FREQUENCY = UAC.MINIMUM_ALLELE_FREQUENCY;
         REPORT_SLOD = ! UAC.NO_SLOD;
         this.verboseWriter = verboseWriter;
-        if ( verboseWriter != null )
-            initializeVerboseWriter(verboseWriter);
         this.beagleWriter = beagleWriter;
-        if ( beagleWriter != null )
-            initializeBeagleWriter(beagleWriter);
-    }
-
-    protected void initializeVerboseWriter(PrintWriter writer) { }
-
-    protected void initializeBeagleWriter(PrintWriter writer) {
-        writer.print("marker alleleA alleleB");
-        for ( String sample : samples ) {
-            writer.print(' ');
-            writer.print(sample);
-        }
-        writer.println();
     }
 
     /**
diff --git a/java/src/org/broadinstitute/sting/gatk/walkers/genotyper/GenotypeCalculationModelFactory.java b/java/src/org/broadinstitute/sting/gatk/walkers/genotyper/GenotypeCalculationModelFactory.java
index 33e6c21e3..5713a7c13 100755
--- a/java/src/org/broadinstitute/sting/gatk/walkers/genotyper/GenotypeCalculationModelFactory.java
+++ b/java/src/org/broadinstitute/sting/gatk/walkers/genotyper/GenotypeCalculationModelFactory.java
@@ -30,7 +30,7 @@ import org.broadinstitute.sting.utils.genotype.GenotypeWriterFactory;
 import org.apache.log4j.Logger;
 
 import java.util.Set;
-import java.io.PrintWriter;
+import java.io.PrintStream;
 
 
 public class GenotypeCalculationModelFactory {
@@ -60,8 +60,8 @@ public class GenotypeCalculationModelFactory {
                                                                    Logger logger,
                                                                    UnifiedArgumentCollection UAC,
                                                                    GenotypeWriterFactory.GENOTYPE_FORMAT outputFormat,
-                                                                   PrintWriter verboseWriter,
-                                                                   PrintWriter beagleWriter) {
+                                                                   PrintStream verboseWriter,
+                                                                   PrintStream beagleWriter) {
         GenotypeCalculationModel gcm;
         switch ( UAC.genotypeModel ) {
             case EM_POINT_ESTIMATE:
diff --git a/java/src/org/broadinstitute/sting/gatk/walkers/genotyper/JointEstimateGenotypeCalculationModel.java b/java/src/org/broadinstitute/sting/gatk/walkers/genotyper/JointEstimateGenotypeCalculationModel.java
index a7a5297d5..5d5e8bc16 100644
--- a/java/src/org/broadinstitute/sting/gatk/walkers/genotyper/JointEstimateGenotypeCalculationModel.java
+++ b/java/src/org/broadinstitute/sting/gatk/walkers/genotyper/JointEstimateGenotypeCalculationModel.java
@@ -107,16 +107,6 @@ public abstract class JointEstimateGenotypeCalculationModel extends GenotypeCalc
         }
     }
 
-    protected void initializeVerboseWriter(PrintWriter verboseWriter) {
-        StringBuilder header = new StringBuilder("AFINFO\tLOC\tMAF\tF\tNullAFpriors\t");
-        for ( char altAllele : BaseUtils.BASES ) {
-            char base = Character.toLowerCase(altAllele);
-            header.append("POfDGivenAFFor" + base + "\t");
-            header.append("PosteriorAFFor" + base + "\t");
-        }
-        verboseWriter.println(header);
-    }
-
     protected void initialize(char ref, Map<String, StratifiedAlignmentContext> contexts, StratifiedAlignmentContext.StratifiedContextType contextType) {
         // by default, no initialization is done
         return;
diff --git a/java/src/org/broadinstitute/sting/gatk/walkers/genotyper/UnifiedGenotyper.java b/java/src/org/broadinstitute/sting/gatk/walkers/genotyper/UnifiedGenotyper.java
index 8d2da6302..fc54b7fe1 100755
--- a/java/src/org/broadinstitute/sting/gatk/walkers/genotyper/UnifiedGenotyper.java
+++ b/java/src/org/broadinstitute/sting/gatk/walkers/genotyper/UnifiedGenotyper.java
@@ -39,8 +39,7 @@ import org.broadinstitute.sting.utils.genotype.glf.GLFGenotypeWriter;
 import org.broadinstitute.sting.utils.genotype.vcf.*;
 
 import java.util.*;
-import java.io.PrintWriter;
-import java.io.FileNotFoundException;
+import java.io.PrintStream;
 
 
 /**
@@ -57,15 +56,10 @@ public class UnifiedGenotyper extends LocusWalker<Pair<VariationCall, List<Genot
     public GenotypeWriter writer = null;
 
     @Argument(fullName = "verbose_mode", shortName = "verbose", doc = "File to print all of the annotated and detailed debugging output", required = false)
-    public String VERBOSE = null;
+    public PrintStream verboseWriter = null;
 
     @Argument(fullName = "beagle_file", shortName = "beagle", doc = "File to print BEAGLE-specific data for use with imputation", required = false)
-    public String BEAGLE = null;
-
-
-    // the verbose and beagle writers
-    private PrintWriter verboseWriter = null;
-    private PrintWriter beagleWriter = null;
+    public PrintStream beagleWriter = null;
 
     // the model used for calculating genotypes
     private ThreadLocal<GenotypeCalculationModel> gcm = new ThreadLocal<GenotypeCalculationModel>();
@@ -109,7 +103,7 @@ public class UnifiedGenotyper extends LocusWalker<Pair<VariationCall, List<Genot
             sb.append("\n***\tUse Q" + (10.0 * UAC.LOD_THRESHOLD) + " as an approximate equivalent to your LOD " + UAC.LOD_THRESHOLD + " cutoff");
             throw new IllegalArgumentException(sb.toString());
         }
-        if ( BEAGLE != null && UAC.genotypeModel == GenotypeCalculationModel.Model.EM_POINT_ESTIMATE ) {
+        if ( beagleWriter != null && UAC.genotypeModel == GenotypeCalculationModel.Model.EM_POINT_ESTIMATE ) {
             throw new IllegalArgumentException("BEAGLE output is not currently supported in the EM_POINT_ESTIMATE calculation model.");
         }
 
@@ -118,11 +112,6 @@ public class UnifiedGenotyper extends LocusWalker<Pair<VariationCall, List<Genot
             // no ASSUME_SINGLE_SAMPLE because the IO system doesn't know how to get the sample name
             if ( UAC.ASSUME_SINGLE_SAMPLE != null )
                 throw new IllegalArgumentException("For technical reasons, the ASSUME_SINGLE_SAMPLE argument cannot be used with multiple threads");
-
-            // TODO -- it would be nice to be able to handle verbose and beagle even with multiple threads
-            // no VERBOSE because we'd need to deal with parallelizing the writing
-            if ( VERBOSE != null || BEAGLE != null )
-                throw new IllegalArgumentException("For technical reasons, the VERBOSE and BEAGLE arguments cannot be used with multiple threads");
         }
 
         // get all of the unique sample names - unless we're in POOLED mode, in which case we ignore the sample names
@@ -149,13 +138,24 @@ public class UnifiedGenotyper extends LocusWalker<Pair<VariationCall, List<Genot
         }
 
         // initialize the writers
-        try {
-            if ( VERBOSE != null )
-                verboseWriter = new PrintWriter(VERBOSE);
-            if ( BEAGLE != null )
-                beagleWriter = new PrintWriter(BEAGLE);
-        } catch (FileNotFoundException e) {
-            throw new StingException("UnifiedGenotyper [verbose/beagle]: could not open file for writing");
+        if ( verboseWriter != null ) {
+            if(UAC.genotypeModel != GenotypeCalculationModel.Model.EM_POINT_ESTIMATE) {
+                StringBuilder header = new StringBuilder("AFINFO\tLOC\tMAF\tF\tNullAFpriors\t");
+                for ( char altAllele : BaseUtils.BASES ) {
+                    char base = Character.toLowerCase(altAllele);
+                    header.append("POfDGivenAFFor" + base + "\t");
+                    header.append("PosteriorAFFor" + base + "\t");
+                }
+                verboseWriter.println(header);
+            }
+        }
+        if ( beagleWriter != null ) {
+            beagleWriter.print("marker alleleA alleleB");
+            for ( String sample : samples ) {
+                beagleWriter.print(' ');
+                beagleWriter.print(sample);
+            }
+            beagleWriter.println();
         }
 
         // *** If we were called by another walker, then we don't ***
@@ -328,11 +328,6 @@ public class UnifiedGenotyper extends LocusWalker<Pair<VariationCall, List<Genot
 
     // Close any file writers
     public void onTraversalDone(Integer sum) {
-        if ( verboseWriter != null )
-            verboseWriter.close();
-        if ( beagleWriter != null )
-            beagleWriter.close();
-
         logger.info("Processed " + sum + " loci that are callable for SNPs");
     }
 }