JAL-845 SplitFrame for "show product" and after aligning from SplitFrame

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

diff --git a/src/jalview/analysis/AlignmentUtils.java b/src/jalview/analysis/AlignmentUtils.java
index 0441b1d..0ae782e 100644 (file)
--- a/src/jalview/analysis/AlignmentUtils.java
+++ b/src/jalview/analysis/AlignmentUtils.java
@@ -23,6 +23,8 @@ package jalview.analysis;
 import jalview.datamodel.AlignedCodonFrame;
 import jalview.datamodel.AlignmentAnnotation;
 import jalview.datamodel.AlignmentI;
+import jalview.datamodel.SearchResults;
+import jalview.datamodel.Sequence;
 import jalview.datamodel.SequenceI;
 import jalview.schemes.ResidueProperties;
 import jalview.util.MapList;
@@ -636,4 +638,146 @@ public class AlignmentUtils
     }
     return gapsToAdd;
   }
+
+  /**
+   * Returns a list of sequences mapped from the given sequences and aligned
+   * (gapped) in the same way. For example, the cDNA for aligned protein, where
+   * a single gap in protein generates three gaps in cDNA.
+   * 
+   * @param sequences
+   * @param gapCharacter
+   * @param mappings
+   * @return
+   */
+  public static List<SequenceI> getAlignedTranslation(
+          List<SequenceI> sequences, char gapCharacter,
+          Set<AlignedCodonFrame> mappings)
+  {
+    List<SequenceI> alignedSeqs = new ArrayList<SequenceI>();
+
+    for (SequenceI seq : sequences)
+    {
+      List<SequenceI> mapped = getAlignedTranslation(seq, gapCharacter,
+              mappings);
+      alignedSeqs.addAll(mapped);
+    }
+    return alignedSeqs;
+  }
+
+  /**
+   * Returns sequences aligned 'like' the source sequence, as mapped by the
+   * given mappings. Normally we expect zero or one 'mapped' sequences, but this
+   * will support 1-to-many as well.
+   * 
+   * @param seq
+   * @param gapCharacter
+   * @param mappings
+   * @return
+   */
+  protected static List<SequenceI> getAlignedTranslation(SequenceI seq,
+          char gapCharacter, Set<AlignedCodonFrame> mappings)
+  {
+    List<SequenceI> result = new ArrayList<SequenceI>();
+    for (AlignedCodonFrame mapping : mappings)
+    {
+      if (mapping.involvesSequence(seq))
+      {
+        SequenceI mapped = getAlignedTranslation(seq, gapCharacter, mapping);
+        if (mapped != null)
+        {
+          result.add(mapped);
+        }
+      }
+    }
+    return result;
+  }
+
+  /**
+   * Returns the translation of 'seq' (as held in the mapping) with
+   * corresponding alignment (gaps).
+   * 
+   * @param seq
+   * @param gapCharacter
+   * @param mapping
+   * @return
+   */
+  protected static SequenceI getAlignedTranslation(SequenceI seq,
+          char gapCharacter, AlignedCodonFrame mapping)
+  {
+    String gap = String.valueOf(gapCharacter);
+    boolean toDna = false;
+    int fromRatio = 1;
+    SequenceI mapTo = mapping.getDnaForAaSeq(seq);
+    if (mapTo != null)
+    {
+      // mapping is from protein to nucleotide
+      toDna = true;
+      // should ideally get gap count ratio from mapping
+      gap = String.valueOf(new char[]
+      { gapCharacter, gapCharacter, gapCharacter });
+    }
+    else
+    {
+      // mapping is from nucleotide to protein
+      mapTo = mapping.getAaForDnaSeq(seq);
+      fromRatio = 3;
+    }
+    StringBuilder newseq = new StringBuilder(seq.getLength()
+            * (toDna ? 3 : 1));
+
+    int residueNo = 0; // in seq, base 1
+    int[] phrase = new int[fromRatio];
+    int phraseOffset = 0;
+    int gapWidth = 0;
+    boolean first = true;
+    final Sequence alignedSeq = new Sequence("", "");
+
+    for (char c : seq.getSequence())
+    {
+      if (c == gapCharacter)
+      {
+        gapWidth++;
+        if (gapWidth >= fromRatio)
+        {
+          newseq.append(gap);
+          gapWidth = 0;
+        }
+      }
+      else
+      {
+        phrase[phraseOffset++] = residueNo + 1;
+        if (phraseOffset == fromRatio)
+        {
+          /*
+           * Have read a whole codon (or protein residue), now translate: map
+           * source phrase to positions in target sequence add characters at
+           * these positions to newseq Note mapping positions are base 1, our
+           * sequence positions base 0.
+           */
+          SearchResults sr = new SearchResults();
+          for (int pos : phrase)
+          {
+            mapping.markMappedRegion(seq, pos, sr);
+          }
+          newseq.append(sr.toString());
+          if (first)
+          {
+            first = false;
+            // Hack: Copy sequence dataset, name and description from
+            // SearchResults.match[0].sequence
+            // TODO? carry over sequence names from original 'complement'
+            // alignment
+            SequenceI mappedTo = sr.getResultSequence(0);
+            alignedSeq.setName(mappedTo.getName());
+            alignedSeq.setDescription(mappedTo.getDescription());
+            alignedSeq.setDatasetSequence(mappedTo);
+          }
+          phraseOffset = 0;
+        }
+        residueNo++;
+      }
+    }
+    alignedSeq.setSequence(newseq.toString());
+    return alignedSeq;
+  }
 }