From 9818c69df67fe5c930df4b62d895b5b9b15fac07 Mon Sep 17 00:00:00 2001
From: Ryan Poplin <rpoplin@broadinstitute.org>
Date: Wed, 25 Jan 2012 09:32:52 -0500
Subject: [PATCH] Can now specify active regions to process at the command
 line, mainly for debugging purposes

---
 .../traversals/TraverseActiveRegions.java     |  4 +--
 .../gatk/walkers/ActiveRegionWalker.java      | 32 +++++++++++++++++++
 2 files changed, 34 insertions(+), 2 deletions(-)

diff --git a/public/java/src/org/broadinstitute/sting/gatk/traversals/TraverseActiveRegions.java b/public/java/src/org/broadinstitute/sting/gatk/traversals/TraverseActiveRegions.java
index ebfcc0c29..cf15cc92b 100644
--- a/public/java/src/org/broadinstitute/sting/gatk/traversals/TraverseActiveRegions.java
+++ b/public/java/src/org/broadinstitute/sting/gatk/traversals/TraverseActiveRegions.java
@@ -92,7 +92,7 @@ public class TraverseActiveRegions <M,T> extends TraversalEngine<M,T,ActiveRegio
 
                 // Call the walkers isActive function for this locus and add them to the list to be integrated later
                 if( initialIntervals.overlaps(location) ) {
-                    final boolean isActive = walker.isActive( tracker, refContext, locus );
+                    final boolean isActive = ( walker.presetActiveRegions == null ? walker.isActive( tracker, refContext, locus ) : walker.presetActiveRegions.overlaps(location) );
                     isActiveList.add( new ActiveRegion(location, isActive, engine.getGenomeLocParser(), activeRegionExtension ) );
                 }
                 
@@ -109,7 +109,7 @@ public class TraverseActiveRegions <M,T> extends TraversalEngine<M,T,ActiveRegio
                 if( !locusView.hasNext() ) {
                     // Call the walkers isActive function for this locus and add them to the list to be integrated later
                     if( initialIntervals.overlaps(location) ) {
-                        final boolean isActive = walker.isActive( tracker, refContext, locus );
+                        final boolean isActive = ( walker.presetActiveRegions == null ? walker.isActive( tracker, refContext, locus ) : walker.presetActiveRegions.overlaps(location) );
                         isActiveList.add( new ActiveRegion(location, isActive, engine.getGenomeLocParser(), activeRegionExtension ) );
                     }
 
diff --git a/public/java/src/org/broadinstitute/sting/gatk/walkers/ActiveRegionWalker.java b/public/java/src/org/broadinstitute/sting/gatk/walkers/ActiveRegionWalker.java
index d7e170d73..508aebb5c 100644
--- a/public/java/src/org/broadinstitute/sting/gatk/walkers/ActiveRegionWalker.java
+++ b/public/java/src/org/broadinstitute/sting/gatk/walkers/ActiveRegionWalker.java
@@ -1,6 +1,11 @@
 package org.broadinstitute.sting.gatk.walkers;
 
 import net.sf.picard.reference.IndexedFastaSequenceFile;
+import org.broad.tribble.Feature;
+import org.broadinstitute.sting.commandline.Input;
+import org.broadinstitute.sting.commandline.IntervalBinding;
+import org.broadinstitute.sting.commandline.Output;
+import org.broadinstitute.sting.commandline.RodBinding;
 import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
 import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
 import org.broadinstitute.sting.gatk.filters.DuplicateReadFilter;
@@ -14,8 +19,10 @@ import org.broadinstitute.sting.utils.GenomeLocParser;
 import org.broadinstitute.sting.utils.GenomeLocSortedSet;
 import org.broadinstitute.sting.utils.activeregion.ActiveRegion;
 import org.broadinstitute.sting.utils.interval.IntervalMergingRule;
+import org.broadinstitute.sting.utils.interval.IntervalSetRule;
 import org.broadinstitute.sting.utils.interval.IntervalUtils;
 
+import java.io.PrintStream;
 import java.util.ArrayList;
 import java.util.List;
 
@@ -32,6 +39,31 @@ import java.util.List;
 @ReadFilters({UnmappedReadFilter.class, NotPrimaryAlignmentFilter.class, DuplicateReadFilter.class, FailsVendorQualityCheckFilter.class})
 public abstract class ActiveRegionWalker<MapType, ReduceType> extends Walker<MapType, ReduceType> {
 
+    @Output(fullName="activeRegionOut", shortName="ARO", doc="Output the active region to this interval list file", required = false)
+    protected PrintStream activeRegionOutStream = null;
+
+    @Input(fullName="activeRegionIn", shortName="AR", doc="Use this interval list file as the active regions to process", required = false)
+    protected List<IntervalBinding<Feature>> activeRegionBindings = null;
+
+    public GenomeLocSortedSet presetActiveRegions = null;
+
+    @Override
+    public void initialize() {
+        if( activeRegionBindings == null ) { return; }
+        List<GenomeLoc> allIntervals = new ArrayList<GenomeLoc>(0);
+        for ( IntervalBinding intervalBinding : activeRegionBindings ) {
+            List<GenomeLoc> intervals = intervalBinding.getIntervals(this.getToolkit());
+
+            if ( intervals.isEmpty() ) {
+                logger.warn("The interval file " + intervalBinding.getSource() + " contains no intervals that could be parsed.");
+            }
+
+            allIntervals = IntervalUtils.mergeListsBySetOperator(intervals, allIntervals, IntervalSetRule.UNION);
+        }
+
+        presetActiveRegions = IntervalUtils.sortAndMergeIntervals(this.getToolkit().getGenomeLocParser(), allIntervals, IntervalMergingRule.ALL);
+    }
+
     // Do we actually want to operate on the context?
     public boolean filter(final RefMetaDataTracker tracker, final ReferenceContext ref, final AlignmentContext context) {
         return true;    // We are keeping all the reads