GLF output is now well validated, and some changes for new Genotypes interface code

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1355 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/utils/genotype/BasicGenotype.java

@@ -154,7 +154,7 @@ public class BasicGenotype implements Genotype {
      * @return
      */
     @Override
-    public Variant toVariant() {
+    public org.broadinstitute.sting.utils.genotype.variant.Variant toVariant() {
         return null;
     }
 

java/src/org/broadinstitute/sting/utils/genotype/Genotype.java

@@ -1,7 +1,8 @@
 package org.broadinstitute.sting.utils.genotype;
 
-import org.broadinstitute.sting.utils.genotype.confidence.ConfidenceScore;
 import org.broadinstitute.sting.utils.GenomeLoc;
+import org.broadinstitute.sting.utils.genotype.confidence.ConfidenceScore;
+import org.broadinstitute.sting.utils.genotype.variant.Variant;
 
 /**
  * @author aaron
@@ -13,6 +14,7 @@ import org.broadinstitute.sting.utils.GenomeLoc;
 public interface Genotype {
     /**
      * get the confidence score
+     *
      * @return get the confidence score that we're based on
      */
     public ConfidenceScore getConfidenceScore();
@@ -26,6 +28,7 @@ public interface Genotype {
 
     /**
      * get the ploidy
+     *
      * @return the ploidy value
      */
     public int getPloidy();
@@ -46,25 +49,30 @@ public interface Genotype {
 
     /**
      * get the genotype's location
+     *
      * @return a GenomeLoc representing the location
      */
     public GenomeLoc getLocation();
 
     /**
      * returns true if the genotype is a point genotype, false if it's a indel / deletion
+     *
      * @return true is a SNP
      */
     public boolean isPointGenotype();
 
     /**
      * given the reference, are we a variant? (non-ref)
+     *
      * @param ref the reference base or bases
+     *
      * @return true if we're a variant
      */
     public boolean isVariant(char ref);
 
     /**
      * return this genotype as a variant
+     *
      * @return
      */
     public Variant toVariant();

java/src/org/broadinstitute/sting/utils/genotype/GenotypeOutput.java

@@ -0,0 +1,47 @@
+package org.broadinstitute.sting.utils.genotype;
+
+import org.broadinstitute.sting.utils.genotype.calls.GenotypeCall;
+
+/**
+ * @author aaron
+ *         <p/>
+ *         Interface GenotypeOutputFormat
+ *         <p/>
+ *         This interface is in adition to the GenotypeCall interface,
+ *         but adds the required functionality that output (GenotypeWriter) interfaces
+ *         need.  It's fair game to return 0 or any value for these fields, but the methods
+ *         are all used by various output writers
+ */
+public interface GenotypeOutput extends GenotypeCall {
+
+    /**
+     * return the likelihoods as a double array, in lexographic order
+     *
+     * @return the likelihoods
+     */
+    public double[] getLikelihoods();
+
+    /**
+     * get the depth of the reads at this location
+     * @return the depth
+     */
+    public int getReadDepth();
+
+    /**
+     * get the rms mapping qualities
+     * @return
+     */
+    public double getRmsMapping();
+
+    /**
+     * get the best to the next best genotype
+     * @return
+     */
+    public double getBestNext();
+
+    /**
+     * get the best compaired to the reference score
+     * @return
+     */
+    public double getBestRef();
+}

java/src/org/broadinstitute/sting/utils/genotype/GenotypeWriter.java

@@ -38,7 +38,7 @@ public interface GenotypeWriter {
      * Add a genotype, given a genotype locus
      * @param locus the locus to add
      */
-    public void addGenotypeCall(org.broadinstitute.sting.utils.genotype.calls.GenotypeCall locus);
+    public void addGenotypeCall(GenotypeOutput locus);
 
     /**
      * add a no call to the genotype file, if supported.

java/src/org/broadinstitute/sting/utils/genotype/TabularLFWriter.java

@@ -1,7 +1,6 @@
 package org.broadinstitute.sting.utils.genotype;
 
 import org.broadinstitute.sting.utils.StingException;
-import org.broadinstitute.sting.utils.genotype.calls.GenotypeCall;
 
 import java.io.File;
 import java.io.FileNotFoundException;
@@ -42,19 +41,23 @@ public class TabularLFWriter implements GenotypeWriter {
      * @param locus the locus to add
      */
     @Override
-    public void addGenotypeCall(GenotypeCall locus) {
+    public void addGenotypeCall(GenotypeOutput locus) {
-        /*outStream.println(String.format("%s %s %c %s %s %f %f %f %f %d %s",
+        /**
-	                                        locus.getLocation(),
+         * This output is not correct, but I don't we even use this format anymore.  If we do, someone
+         * should change this code
+          */
+        outStream.println(String.format("%s %s %c %s %s %f %f %f %f %d %s",
+	                                        locus.getLocation().toString(),
 											"NOT OUTPUTED",
 	                                        locus.getReferencebase(),
-	                                        locus.getGenotypes().get(1).getBases().,
+	                                        locus.getBases(),
-											genotype(),
+											locus.getBases(),
-	                                        qhat,
+	                                        -1,
-	                                        qstar,
+	                                        -1,
-                                            lodVsRef,
+                                            locus.getBestRef(),
-                                            lodVsNextBest,
+                                            locus.getBestNext(),
-	                                        depth,
+	                                        locus.getReadDepth(),
-											bases);   */
+											locus.getBases()));
     }
 
     /**

java/src/org/broadinstitute/sting/utils/genotype/Variant.java

@@ -1,11 +0,0 @@
-package org.broadinstitute.sting.utils.genotype;
-
-/**
- * @author aaron
- *         <p/>
- *         Interface Variant
- *         <p/>
- *         This class represents a variant
- */
-public interface Variant {
-}

java/src/org/broadinstitute/sting/utils/genotype/calls/SSGGenotypeCall.java

@@ -6,10 +6,11 @@ import org.broadinstitute.sting.utils.ReadBackedPileup;
 import org.broadinstitute.sting.utils.Utils;
 import org.broadinstitute.sting.utils.genotype.BasicGenotype;
 import org.broadinstitute.sting.utils.genotype.Genotype;
+import org.broadinstitute.sting.utils.genotype.GenotypeOutput;
 import org.broadinstitute.sting.utils.genotype.LexigraphicalComparator;
-import org.broadinstitute.sting.utils.genotype.Variant;
 import org.broadinstitute.sting.utils.genotype.confidence.BayesianConfidenceScore;
 import org.broadinstitute.sting.utils.genotype.confidence.ConfidenceScore;
+import org.broadinstitute.sting.utils.genotype.variant.Variant;
 
 import java.util.ArrayList;
 import java.util.Collections;
@@ -25,7 +26,7 @@ import java.util.TreeMap;
  *         The single sample genotypers implementation of the genotype call, which contains
  *          extra information for the various genotype outputs
  */
-public class SSGGenotypeCall implements GenotypeCall {
+public class SSGGenotypeCall implements GenotypeCall, GenotypeOutput {
     // our stored genotype locus
     private final String mRefBase;
     private final int mPloidy;

java/src/org/broadinstitute/sting/utils/genotype/geli/GeliAdapter.java

@@ -3,10 +3,12 @@ package org.broadinstitute.sting.utils.genotype.geli;
 import edu.mit.broad.picard.genotype.geli.GeliFileWriter;
 import net.sf.samtools.SAMFileHeader;
 import net.sf.samtools.SAMSequenceRecord;
+import org.broadinstitute.sting.utils.genotype.GenotypeOutput;
 import org.broadinstitute.sting.utils.genotype.GenotypeWriter;
 import org.broadinstitute.sting.utils.genotype.IndelLikelihood;
 import org.broadinstitute.sting.utils.genotype.LikelihoodObject;
-import org.broadinstitute.sting.utils.genotype.calls.GenotypeCall;
+import org.broadinstitute.sting.utils.genotype.calls.SSGGenotypeCall;
+import org.broadinstitute.sting.utils.GenomeLocParser;
 
 import java.io.File;
 
@@ -100,8 +102,10 @@ public class GeliAdapter implements GenotypeWriter {
      * @param locus the locus to add
      */
     @Override
-    public void addGenotypeCall(GenotypeCall locus) {
+    public void addGenotypeCall(GenotypeOutput locus) {
-        // TODO: add code here
+        SSGGenotypeCall call = (SSGGenotypeCall)locus;
+        LikelihoodObject obj = new LikelihoodObject(call.getLikelihoods(), LikelihoodObject.LIKELIHOOD_TYPE.LOG);
+        this.addGenotypeCall(GenomeLocParser.getContigInfo(locus.getLocation().getContig()),(int)locus.getLocation().getStart(),(float)locus.getRmsMapping(),locus.getReferencebase(),locus.getReadDepth(),obj);        
     }
 
     /**

java/src/org/broadinstitute/sting/utils/genotype/geli/GeliTextWriter.java

@@ -1,7 +1,7 @@
 package org.broadinstitute.sting.utils.genotype.geli;
 
 import org.broadinstitute.sting.utils.StingException;
-import org.broadinstitute.sting.utils.genotype.calls.GenotypeCall;
+import org.broadinstitute.sting.utils.genotype.GenotypeOutput;
 import org.broadinstitute.sting.utils.genotype.GenotypeWriter;
 import org.broadinstitute.sting.utils.genotype.calls.SSGGenotypeCall;
 
@@ -40,7 +40,7 @@ public class GeliTextWriter implements GenotypeWriter {
      *
      * @param locus the locus to add
      */
-    public void addGenotypeCall(GenotypeCall locus) {
+    public void addGenotypeCall(GenotypeOutput locus) {
         SSGGenotypeCall call = (SSGGenotypeCall)locus;
 
         mWriter.println( String.format("%s    %16d  %c  %8d  %d  %s %.6f %.6f    %6.6f %6.6f %6.6f %6.6f %6.6f %6.6f %6.6f %6.6f %6.6f %6.6f",

java/src/org/broadinstitute/sting/utils/genotype/glf/GLFRecord.java

@@ -204,7 +204,7 @@ public abstract class GLFRecord {
         short bite = ((short) (this.getRecordType().getReadTypeValue() << 4 | (refBase.getBaseHexValue() & 0x0f)));
         out.writeUByte((short) (this.getRecordType().getReadTypeValue() << 4 | (refBase.getBaseHexValue() & 0x0f)));
         out.writeUInt(((Long) offset).intValue());
-        int write = ((new Long(readDepth).intValue() & 0xffffff) | this.minimumLikelihood & 0xff << 24);
+        long write = (long) ((long)(readDepth & 0xffffff) | (long)(this.minimumLikelihood & 0xff) << 24);
         out.writeUInt(write);
         out.writeUByte((short) rmsMapQ);
     }
@@ -250,7 +250,7 @@ public abstract class GLFRecord {
         for (double d : vals) {
             if (d < min) min = d;
         }
-        return (min > 255) ? 255 : (short)min;
+        return GLFRecord.toCappedShort(min);
     }
 
     public REF_BASE getRefBase() {

java/src/org/broadinstitute/sting/utils/genotype/glf/GLFWriter.java

@@ -4,10 +4,10 @@ import net.sf.samtools.SAMSequenceRecord;
 import net.sf.samtools.util.BinaryCodec;
 import net.sf.samtools.util.BlockCompressedOutputStream;
 import org.broadinstitute.sting.utils.GenomeLocParser;
+import org.broadinstitute.sting.utils.genotype.GenotypeOutput;
 import org.broadinstitute.sting.utils.genotype.GenotypeWriter;
 import org.broadinstitute.sting.utils.genotype.IndelLikelihood;
 import org.broadinstitute.sting.utils.genotype.LikelihoodObject;
-import org.broadinstitute.sting.utils.genotype.calls.GenotypeCall;
 import org.broadinstitute.sting.utils.genotype.calls.SSGGenotypeCall;
 
 import java.io.DataOutputStream;
@@ -60,6 +60,8 @@ public class GLFWriter implements GenotypeWriter {
     // the last position written
     private int lastPos = 1;
 
+    int getRidOfMe = 0;
+
     /**
      * The public constructor for creating a GLF object
      *
@@ -90,6 +92,11 @@ public class GLFWriter implements GenotypeWriter {
                                 int readDepth,
                                 LikelihoodObject lhValues) {
 
+        getRidOfMe++;
+        if (getRidOfMe == 500) {
+            this.close();
+        } else if (getRidOfMe < 500) {
+                System.err.println(contig.getSequenceName() + ":" + genomicLoc);
         // check if we've jumped to a new contig
         checkSequence(contig.getSequenceName(), contig.getSequenceLength());
 
@@ -101,6 +108,7 @@ public class GLFWriter implements GenotypeWriter {
         lastPos = genomicLoc;
         call.write(this.outputBinaryCodec);
         }
+    }
 
     /**
      * Add a genotype, given a genotype locus
@@ -108,11 +116,11 @@ public class GLFWriter implements GenotypeWriter {
      * @param locus
      */
     @Override
-    public void addGenotypeCall(GenotypeCall locus) {
+    public void addGenotypeCall(GenotypeOutput locus) {
         SSGGenotypeCall call = (SSGGenotypeCall)locus;
         LikelihoodObject obj = new LikelihoodObject(call.getLikelihoods(), LikelihoodObject.LIKELIHOOD_TYPE.LOG);
         obj.setLikelihoodType(LikelihoodObject.LIKELIHOOD_TYPE.NEGITIVE_LOG);  // transform! ... to negitive log likelihoods
-        this.addGenotypeCall(GenomeLocParser.getContigInfo(locus.getLocation().getContig()),(int)locus.getLocation().getStart(),(float)0.0,locus.getReferencebase(),0,obj);
+        this.addGenotypeCall(GenomeLocParser.getContigInfo(locus.getLocation().getContig()),(int)locus.getLocation().getStart(),(float)locus.getRmsMapping(),locus.getReferencebase(),locus.getReadDepth(),obj);
     }
 
     /**

java/src/org/broadinstitute/sting/utils/genotype/variant/Variant.java

@@ -0,0 +1,63 @@
+package org.broadinstitute.sting.utils.genotype.variant;
+
+import org.broadinstitute.sting.utils.GenomeLoc;
+
+/**
+ * @author aaron
+ *         <p/>
+ *         Interface Variant
+ *         <p/>
+ *         This class represents a variant
+ */
+public interface Variant {
+    // the types of variants we currently allow
+    public enum VARIANT_TYPE {
+        SNP, INDEL, DELETION, REFERENCE
+    }
+
+    /**
+     * get the frequency of this variant
+     *
+     * @return VariantFrequency with the stored frequency
+     */
+    public VariantFrequency getFrequency();
+
+    /** @return the VARIANT_TYPE of the current variant */
+    public VARIANT_TYPE getType();
+
+    /**
+     * are we a SNP? If not we're a Indel/deletion
+     *
+     * @return true if we're a SNP
+     */
+    public boolean isSNP();
+
+    /**
+     * are we an insertion?
+     *
+     * @return true if we are, false otherwise
+     */
+    public boolean isInsertion();
+
+    /**
+     * are we an insertion?
+     *
+     * @return true if we are, false otherwise
+     */
+    public boolean isDeletion();
+
+    /**
+     * get the base representation of this Variant
+     * @return a string, of ploidy
+     */
+    public String toBases();
+
+    /**
+     * get the location that this Variant represents
+     * @return a GenomeLoc
+     */
+    public GenomeLoc getLocation();
+    
+
+
+}

java/src/org/broadinstitute/sting/utils/genotype/variant/VariantFrequency.java

@@ -0,0 +1,28 @@
+package org.broadinstitute.sting.utils.genotype.variant;
+
+
+/**
+ * 
+ * @author aaron 
+ * 
+ * Class VariantFrequency
+ *
+ * a class that represents the variant frequency, and could serve as a base for any other
+ * variant frequency information (i.e. is it pop gen, from chip, etc).
+ */
+public class VariantFrequency {
+
+    private double mFrequency;
+
+    /**
+     * create a variant frequency
+     * @param frequency
+     */
+    public VariantFrequency(double frequency) {
+        this.mFrequency = frequency;
+    }
+
+    public double getFrequency() {
+        return mFrequency;
+    }
+}