Support for selecting only variants with specific IDs from a file in SelectVariants

-- Cleaned up unused variables as well

public/java/src/org/broadinstitute/sting/gatk/walkers/variantutils/SelectVariants.java

@@ -32,6 +32,7 @@ import org.broadinstitute.sting.utils.codecs.vcf.*;
 import org.broadinstitute.sting.utils.exceptions.UserException;
 import org.broadinstitute.sting.gatk.GenomeAnalysisEngine;
 import org.broadinstitute.sting.utils.MendelianViolation;
+import org.broadinstitute.sting.utils.text.XReadLines;
 import org.broadinstitute.sting.utils.variantcontext.*;
 import org.broadinstitute.sting.commandline.Argument;
 import org.broadinstitute.sting.commandline.Output;
@@ -42,6 +43,7 @@ import org.broadinstitute.sting.gatk.walkers.RodWalker;
 import org.broadinstitute.sting.utils.SampleUtils;
 
 import java.io.File;
+import java.io.FileNotFoundException;
 import java.io.PrintStream;
 import java.util.*;
 
@@ -140,7 +142,7 @@ import java.util.*;
  *   -R ref.fasta \
  *   -T SelectVariants \
  *   --variant input.vcf \
- *   -family NA12891+NA12892=NA12878 \
+ *   -bed family.ped \
  *   -mvq 50 \
  *   -o violations.vcf
  *
@@ -250,16 +252,6 @@ public class SelectVariants extends RodWalker<Integer, Integer> implements TreeR
     @Argument(fullName="keepOriginalAC", shortName="keepOriginalAC", doc="Don't update the AC, AF, or AN values in the INFO field after selecting", required=false)
     private boolean KEEP_ORIGINAL_CHR_COUNTS = false;
 
-    @Hidden
-    @Argument(fullName="family_structure_file", shortName="familyFile", doc="use -family unless you know what you're doing", required=false)
-    private File FAMILY_STRUCTURE_FILE = null;
-
-    /**
-     * String formatted as dad+mom=child where these parameters determine which sample names are examined.
-     */
-    @Argument(fullName="family_structure", shortName="family", doc="string formatted as dad+mom=child where these parameters determine which sample names are examined", required=false)
-    private String FAMILY_STRUCTURE = "";
-
     /**
      * This activates the mendelian violation module that will select all variants that correspond to a mendelian violation following the rules given by the family structure.
      */
@@ -293,6 +285,14 @@ public class SelectVariants extends RodWalker<Integer, Integer> implements TreeR
     @Argument(fullName="selectTypeToInclude", shortName="selectType", doc="Select only a certain type of variants from the input file. Valid types are INDEL, SNP, MIXED, MNP, SYMBOLIC, NO_VARIATION. Can be specified multiple times", required=false)
     private List<VariantContext.Type> TYPES_TO_INCLUDE = new ArrayList<VariantContext.Type>();
 
+    /**
+     * If provided, we will only include variants whose ID field is present in this list of ids.  The matching
+     * is exact string matching.  The file format is just one ID per line
+     *
+     */
+    @Argument(fullName="keepIDs", shortName="IDs", doc="Only emit sites whose ID is found in this file (one ID per line)", required=false)
+    private File rsIDFile = null;
+
 
     @Hidden
     @Argument(fullName="outMVFile", shortName="outMVFile", doc="", required=false)
@@ -339,17 +339,13 @@ public class SelectVariants extends RodWalker<Integer, Integer> implements TreeR
     private int positionToAdd = 0;
     private RandomVariantStructure [] variantArray;
 
-
-    /* Variables used for random selection with AF boosting */
-    private ArrayList<Double> afBreakpoints = null;
-    private ArrayList<Double> afBoosts = null;
-    double bkDelta = 0.0;
-
     private PrintStream outMVFileStream = null;
 
     //Random number generator for the genotypes to remove
     private Random randomGenotypes = new Random();
 
+    private Set<String> IDsToKeep = null;
+
     /**
      * Set up the VCF writer, the sample expressions and regexs, and the JEXL matcher
      */
@@ -437,7 +433,18 @@ public class SelectVariants extends RodWalker<Integer, Integer> implements TreeR
         if (SELECT_RANDOM_FRACTION) logger.info("Selecting approximately " + 100.0*fractionRandom + "% of the variants at random from the variant track");
 
 
-
+        /** load in the IDs file to a hashset for matching */
+        if ( rsIDFile != null ) {
+            IDsToKeep = new HashSet<String>();
+            try {
+                for ( final String line : new XReadLines(rsIDFile).readLines() ) {
+                    IDsToKeep.add(line.trim());
+                }
+                logger.info("Selecting only variants with one of " + IDsToKeep.size() + " IDs from " + rsIDFile);
+            } catch ( FileNotFoundException e ) {
+                throw new UserException.CouldNotReadInputFile(rsIDFile, e);
+            }
+        }
     }
 
     /**
@@ -460,6 +467,9 @@ public class SelectVariants extends RodWalker<Integer, Integer> implements TreeR
         }
 
         for (VariantContext vc : vcs) {
+            if ( IDsToKeep != null && ! IDsToKeep.contains(vc.getID()) )
+                continue;
+
             if (MENDELIAN_VIOLATIONS && mv.countViolations(this.getSampleDB().getFamilies(samples),vc) < 1)
                 break;
 
@@ -513,10 +523,7 @@ public class SelectVariants extends RodWalker<Integer, Integer> implements TreeR
                 else if (!SELECT_RANDOM_FRACTION || ( GenomeAnalysisEngine.getRandomGenerator().nextDouble() < fractionRandom)) {
                     vcfWriter.add(sub);
                 }
-
-
             }
-
         }
 
         return 1;