Potential bug fix for bad references where some codons may have Ns

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4075 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/playground/gatk/walkers/annotator/AminoAcid.java

@@ -82,6 +82,11 @@ public class AminoAcid {
         return "*".equals(getLetter());
     }
 
+    /** Returns true if the amino acid is really just a stop codon. */
+    public boolean isUnknown() {
+        return "X".equals(getLetter());
+    }
+
     public String toString() {
         return name;
     }

java/src/org/broadinstitute/sting/playground/gatk/walkers/annotator/AminoAcidTable.java

@@ -34,6 +34,7 @@ import java.util.HashMap;
 public class AminoAcidTable {
 
 
+    protected static final AminoAcid UNKNOWN = new AminoAcid("X" , "Unknown", "Unk");
     protected static final AminoAcid ISOLEUCINE = new AminoAcid("I" , "Isoleucine", "Ile");
     protected static final AminoAcid LEUCINE = new AminoAcid("L" , "Leucine", "Leu");
     protected static final AminoAcid VALINE =    new AminoAcid("V" , "Valine", "Val");
@@ -177,24 +178,17 @@ public class AminoAcidTable {
         mitochondrialAminoAcidTable.put("TGA", TRYPTOPHAN);
     }
 
-
-
-
     /**
      * Returns the amino acid encoded by the given codon in a eukaryotic genome.
      *
      * @param codon The 3-letter mRNA nucleotide codon 5' to 3'. Expects T's instead of U's. Not case sensitive.
      *
-     * @return The amino acid matching the given codon.
+     * @return The amino acid matching the given codon, or the UNKNOWN amino acid if the codon string doesn't match anything
      */
     public static AminoAcid getEukaryoticAA(String codon) {
         codon = codon.toUpperCase();
         final AminoAcid aa = aminoAcidTable.get(codon);
-        if(aa == null) {
+        return aa == null ? UNKNOWN : aa;
-            throw new IllegalArgumentException("Invalid codon: " + codon);
-        } else {
-            return aa;
-        }
     }
 
 
@@ -204,13 +198,13 @@ public class AminoAcidTable {
      * @param codon The 3-letter mRNA nucleotide codon 5' to 3'. Expects T's instead of U's. Not case sensitive.
      * @param isFirstCodon If this is the 1st codon in the gene, then "ATT" encodes Methyonine
      *
-     * @return The amino acid matching the given codon in mitochondrial genes.
+     * @return The amino acid matching the given codon in mitochondrial genes, or the UNKNOWN amino acid if the codon string doesn't match anything
      */
     public static AminoAcid getMitochondrialAA(String codon, boolean isFirstCodon) {
         codon = codon.toUpperCase();
         final AminoAcid aa = mitochondrialAminoAcidTable.get(codon);
         if(aa == null) {
-            throw new IllegalArgumentException("Invalid codon: " + codon);
+            return UNKNOWN;
         } else if(isFirstCodon && codon.equals("ATT")) {
             return METHIONINE; //special case - 'ATT' in the first codon of a mitochondrial gene codes for methionine instead of isoleucine
         } else {

java/src/org/broadinstitute/sting/playground/gatk/walkers/annotator/TranscriptToInfo.java

@@ -753,17 +753,22 @@ public class TranscriptToInfo extends RodWalker<TreeMap<String, String>, TreeMap
                     else if(positionType == PositionType.CDS)
                     {
                         final String referenceCodon = Character.toString(currentCodon_5to3[0]) + Character.toString(currentCodon_5to3[1]) + currentCodon_5to3[2];
-                        outputLineFields.put(OUTPUT_FRAME, Integer.toString(frame) );
+                        final String variantCodon = Character.toString(currentCodon_5to3[0]) + Character.toString(currentCodon_5to3[1]) + currentCodon_5to3[2];
-                        outputLineFields.put(OUTPUT_CODON_NUMBER, Integer.toString(codonCount_from5) );
 
                         final AminoAcid refAA = isMitochondrialTranscript ? AminoAcidTable.getMitochondrialAA( referenceCodon, codonCount_from5 == 1 ) : AminoAcidTable.getEukaryoticAA( referenceCodon ) ;
+                        final AminoAcid variantAA = isMitochondrialTranscript ? AminoAcidTable.getMitochondrialAA( variantCodon, codonCount_from5 == 1 ) : AminoAcidTable.getEukaryoticAA( variantCodon ) ;
+
+                        if ( refAA.isUnknown() || variantAA.isUnknown() ) {
+                            logger.warn("Illegal amino acid detected: refCodon=" + referenceCodon + " altCodon=" + variantCodon);
+                        }
+
+                        outputLineFields.put(OUTPUT_FRAME, Integer.toString(frame) );
+                        outputLineFields.put(OUTPUT_CODON_NUMBER, Integer.toString(codonCount_from5) );
                         outputLineFields.put(OUTPUT_REFERENCE_CODON, referenceCodon );
                         outputLineFields.put(OUTPUT_REFERENCE_AA, refAA.getCode());
 
                         final char temp = currentCodon_5to3[frame];
                         currentCodon_5to3[frame] = haplotypeAlternate;
-                        final String variantCodon = Character.toString(currentCodon_5to3[0]) + Character.toString(currentCodon_5to3[1]) + currentCodon_5to3[2];
-                        final AminoAcid variantAA = isMitochondrialTranscript ? AminoAcidTable.getMitochondrialAA( variantCodon, codonCount_from5 == 1 ) : AminoAcidTable.getEukaryoticAA( variantCodon ) ;
                         outputLineFields.put(OUTPUT_VARIANT_CODON, variantCodon );
                         outputLineFields.put(OUTPUT_VARIANT_AA, variantAA.getCode());