JAL-3187 update alignAs code and unit tests for 'shared dataset' case

[jalview.git] / src / jalview / analysis / Dna.java

diff --git a/src/jalview/analysis/Dna.java b/src/jalview/analysis/Dna.java
index c3408bd..9611a4c 100644 (file)
--- a/src/jalview/analysis/Dna.java
+++ b/src/jalview/analysis/Dna.java
@@ -44,6 +44,7 @@ import jalview.util.ShiftList;
 import java.util.ArrayList;
 import java.util.Arrays;
 import java.util.Comparator;
+import java.util.Iterator;
 import java.util.List;
 
 public class Dna
@@ -60,7 +61,7 @@ public class Dna
 
   private final String[] seqstring;
 
-  private final List<int[]> contigs;
+  private final Iterator<int[]> contigs;
 
   private final char gapChar;
 
@@ -70,6 +71,10 @@ public class Dna
 
   private final AlignmentI dataset;
 
+  private ShiftList vismapping;
+
+  private int[] startcontigs;
+
   /*
    * Working variables for the translation.
    * 
@@ -91,7 +96,7 @@ public class Dna
    * @param viewport
    * @param visibleContigs
    */
-  public Dna(AlignViewportI viewport, List<int[]> visibleContigs)
+  public Dna(AlignViewportI viewport, Iterator<int[]> visibleContigs)
   {
     this.selection = Arrays.asList(viewport.getSequenceSelection());
     this.seqstring = viewport.getViewAsString(true);
@@ -100,6 +105,45 @@ public class Dna
     this.annotations = viewport.getAlignment().getAlignmentAnnotation();
     this.dnaWidth = viewport.getAlignment().getWidth();
     this.dataset = viewport.getAlignment().getDataset();
+    initContigs();
+  }
+
+  /**
+   * Initialise contigs used as starting point for translateCodingRegion
+   */
+  private void initContigs()
+  {
+    vismapping = new ShiftList(); // map from viscontigs to seqstring
+    // intervals
+
+    int npos = 0;
+    int[] lastregion = null;
+    ArrayList<Integer> tempcontigs = new ArrayList<>();
+    while (contigs.hasNext())
+    {
+      int[] region = contigs.next();
+      if (lastregion == null)
+      {
+        vismapping.addShift(npos, region[0]);
+      }
+      else
+      {
+        // hidden region
+        vismapping.addShift(npos, region[0] - lastregion[1] + 1);
+      }
+      lastregion = region;
+      tempcontigs.add(region[0]);
+      tempcontigs.add(region[1]);
+    }
+
+    startcontigs = new int[tempcontigs.size()];
+    int i = 0;
+    for (Integer val : tempcontigs)
+    {
+      startcontigs[i] = val;
+      i++;
+    }
+    tempcontigs = null;
   }
 
   /**
@@ -150,10 +194,11 @@ public class Dna
   }
 
   /**
+   * Translates cDNA using the specified code table
    * 
    * @return
    */
-  public AlignmentI translateCdna()
+  public AlignmentI translateCdna(GeneticCodeI codeTable)
   {
     AlignedCodonFrame acf = new AlignedCodonFrame();
 
@@ -165,7 +210,7 @@ public class Dna
     for (s = 0; s < sSize; s++)
     {
       SequenceI newseq = translateCodingRegion(selection.get(s),
-              seqstring[s], acf, pepseqs);
+              seqstring[s], acf, pepseqs, codeTable);
 
       if (newseq != null)
       {
@@ -385,37 +430,21 @@ public class Dna
    * @param acf
    *          Definition of global ORF alignment reference frame
    * @param proteinSeqs
+   * @param codeTable
    * @return sequence ready to be added to alignment.
    */
   protected SequenceI translateCodingRegion(SequenceI selection,
           String seqstring, AlignedCodonFrame acf,
-          List<SequenceI> proteinSeqs)
+          List<SequenceI> proteinSeqs, GeneticCodeI codeTable)
   {
     List<int[]> skip = new ArrayList<>();
-    int skipint[] = null;
-    ShiftList vismapping = new ShiftList(); // map from viscontigs to seqstring
-    // intervals
-    int vc = 0;
-    int[] scontigs = new int[contigs.size() * 2];
+    int[] skipint = null;
+
     int npos = 0;
-    int[] lastregion = null;
-    for (int[] region : contigs)
-    {
-      if (lastregion == null)
-      {
-        vismapping.addShift(npos, region[0]);
-      }
-      else
-      {
-        // hidden region
-        vismapping.addShift(npos, region[0] - lastregion[1] + 1);
-      }
-      lastregion = region;
+    int vc = 0;
 
-      scontigs[vc] = region[0];
-      scontigs[vc + 1] = region[1];
-      vc++;
-    }
+    int[] scontigs = new int[startcontigs.length];
+    System.arraycopy(startcontigs, 0, scontigs, 0, startcontigs.length);
 
     // allocate a roughly sized buffer for the protein sequence
     StringBuilder protein = new StringBuilder(seqstring.length() / 2);
@@ -439,9 +468,8 @@ public class Dna
         /*
          * Filled up a reading frame...
          */
-        AlignedCodon alignedCodon = new AlignedCodon(cdp[0], cdp[1],
-                cdp[2]);
-        String aa = ResidueProperties.codonTranslate(new String(codon));
+        AlignedCodon alignedCodon = new AlignedCodon(cdp[0], cdp[1], cdp[2]);
+        String aa = codeTable.translate(new String(codon));
         rf = 0;
         final String gapString = String.valueOf(gapChar);
         if (aa == null)
@@ -548,7 +576,7 @@ public class Dna
             skip.add(skipint);
             skipint = null;
           }
-          if (aa.equals("STOP"))
+          if (aa.equals(ResidueProperties.STOP))
           {
             aa = STOP_ASTERIX;
           }
@@ -855,6 +883,23 @@ public class Dna
   }
 
   /**
+   * Answers the reverse complement of the input string
+   * 
+   * @see #getComplement(char)
+   * @param s
+   * @return
+   */
+  public static String reverseComplement(String s)
+  {
+    StringBuilder sb = new StringBuilder(s.length());
+    for (int i = s.length() - 1; i >= 0; i--)
+    {
+      sb.append(Dna.getComplement(s.charAt(i)));
+    }
+    return sb.toString();
+  }
+
+  /**
    * Returns dna complement (preserving case) for aAcCgGtTuU. Ambiguity codes
    * are treated as on http://reverse-complement.com/. Anything else is left
    * unchanged.