JAL-2110 added alignAsSameSequences to handle 'direct' (copy) case

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

diff --git a/src/jalview/analysis/AlignmentUtils.java b/src/jalview/analysis/AlignmentUtils.java
index 20fcf25..17aab15 100644 (file)
--- a/src/jalview/analysis/AlignmentUtils.java
+++ b/src/jalview/analysis/AlignmentUtils.java
@@ -2535,6 +2535,17 @@ public class AlignmentUtils
    */
   public static int alignAs(AlignmentI unaligned, AlignmentI aligned)
   {
+    /*
+     * easy case - aligning a copy of aligned sequences
+     */
+    if (alignAsSameSequences(unaligned, aligned))
+    {
+      return unaligned.getHeight();
+    }
+
+    /*
+     * fancy case - aligning via mappings between sequences
+     */
     List<SequenceI> unmapped = new ArrayList<SequenceI>();
     Map<Integer, Map<SequenceI, Character>> columnMap = buildMappedColumnsMap(
             unaligned, aligned, unmapped);
@@ -2587,6 +2598,54 @@ public class AlignmentUtils
   }
 
   /**
+   * If unaligned and aligned sequences share the same dataset sequences, then
+   * simply copies the aligned sequences to the unaligned sequences and returns
+   * true; else returns false
+   * 
+   * @param unaligned
+   * @param aligned
+   * @return
+   */
+  static boolean alignAsSameSequences(AlignmentI unaligned,
+          AlignmentI aligned)
+  {
+    if (aligned.getDataset() == null || unaligned.getDataset() == null)
+    {
+      return false; // should only pass alignments with datasets here
+    }
+
+    Map<SequenceI, SequenceI> alignedDatasets = new HashMap<SequenceI, SequenceI>();
+    for (SequenceI seq : aligned.getSequences())
+    {
+      alignedDatasets.put(seq.getDatasetSequence(), seq);
+    }
+
+    /*
+     * first pass - check whether all sequences to be aligned share a dataset
+     * sequence with an aligned sequence
+     */
+    for (SequenceI seq : unaligned.getSequences())
+    {
+      if (!alignedDatasets.containsKey(seq.getDatasetSequence()))
+      {
+        return false;
+      }
+    }
+
+    /*
+     * second pass - copy aligned sequences
+     */
+    for (SequenceI seq : unaligned.getSequences())
+    {
+      SequenceI alignedSequence = alignedDatasets.get(seq
+              .getDatasetSequence());
+      seq.setSequence(alignedSequence.getSequenceAsString());
+    }
+
+    return true;
+  }
+
+  /**
    * Returns a map whose key is alignment column number (base 1), and whose
    * values are a map of sequence characters in that column.
    *