Merge branch 'docs/2_8_1_Release' into Release_2_8_2_Branch

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

diff --git a/src/jalview/analysis/AlignSeq.java b/src/jalview/analysis/AlignSeq.java
index 5b37934..8147d76 100755 (executable)
--- a/src/jalview/analysis/AlignSeq.java
+++ b/src/jalview/analysis/AlignSeq.java
@@ -17,7 +17,6 @@
  * The Jalview Authors are detailed in the 'AUTHORS' file.
  */
 package jalview.analysis;
-
 import java.util.*;
 
 import java.awt.*;
@@ -268,16 +267,11 @@ public class AlignSeq
   }
 
   /**
-   * DOCUMENT ME!
-   * 
-   * @param s1
-   *          DOCUMENT ME!
-   * @param string1
-   *          - string to align for sequence1
-   * @param s2
-   *          sequence 2
-   * @param string2
-   *          - string to align for sequence2
+   * Construct score matrix for sequences with standard DNA or PEPTIDE matrix
+   * @param s1 - sequence 1
+   * @param string1 - string to use for s1
+   * @param s2 - sequence 2
+   * @param string2 - string to use for s2
    * @param type
    *          DNA or PEPTIDE
    */
@@ -290,6 +284,14 @@ public class AlignSeq
     SeqInit(string1, string2);
   }
 
+  /**
+   * Construct score matrix for sequences with custom substitution matrix
+   * @param s1 - sequence 1
+   * @param string1 - string to use for s1
+   * @param s2 - sequence 2
+   * @param string2 - string to use for s2
+   * @param scoreMatrix - substitution matrix to use for alignment
+   */
   public void SeqInit(SequenceI s1, String string1, SequenceI s2,
           String string2, ScoreMatrix scoreMatrix)
   {
@@ -303,7 +305,7 @@ public class AlignSeq
    * construct score matrix for string1 and string2 (after removing any existing
    * gaps
    * 
-   * @param string1
+   * @param string1 
    * @param string2
    */
   private void SeqInit(String string1, String string2)
@@ -921,6 +923,94 @@ public class AlignSeq
   }
 
   /**
+   * Compute a globally optimal needleman and wunsch alignment between two
+   * sequences
+   * 
+   * @param s1
+   * @param s2
+   * @param type
+   *          AlignSeq.DNA or AlignSeq.PEP
+   */
+  public static AlignSeq doGlobalNWAlignment(SequenceI s1, SequenceI s2,
+          String type)
+  {
+    AlignSeq as = new AlignSeq(s1, s2, type);
+
+    as.calcScoreMatrix();
+    as.traceAlignment();
+    return as;
+  }
+
+  /**
+   * 
+   * @return mapping from positions in S1 to corresponding positions in S2
+   */
+  public jalview.datamodel.Mapping getMappingFromS1(boolean allowmismatch)
+  {
+    ArrayList<Integer> as1 = new ArrayList<Integer>(), as2 = new ArrayList<Integer>();
+    int pdbpos = s2.getStart() + getSeq2Start() - 2;
+    int alignpos = s1.getStart() + getSeq1Start() - 2;
+    int lp2 = pdbpos - 3, lp1 = alignpos - 3;
+    boolean lastmatch = false;
+    // and now trace the alignment onto the atom set.
+    for (int i = 0; i < astr1.length(); i++)
+    {
+      char c1 = astr1.charAt(i), c2 = astr2.charAt(i);
+      if (c1 != '-')
+      {
+        alignpos++;
+      }
+
+      if (c2 != '-')
+      {
+        pdbpos++;
+      }
+
+      if (allowmismatch || c1 == c2)
+      {
+        lastmatch = true;
+        // extend mapping interval.
+        if (lp1 + 1 != alignpos || lp2+1 !=pdbpos)
+        {
+          as1.add(Integer.valueOf(alignpos));
+          as2.add(Integer.valueOf(pdbpos));
+        }
+        lp1 = alignpos;
+        lp2 = pdbpos;
+      }
+      else
+      {
+        lastmatch = false;
+      }
+    }
+    // construct range pairs
+    int[] mapseq1 = new int[as1.size() + (lastmatch ? 1 : 0)], mapseq2 = new int[as2
+            .size() + (lastmatch ? 1 : 0)];
+    int i = 0;
+    for (Integer ip : as1)
+    {
+      mapseq1[i++] = ip;
+    }
+    ;
+    i = 0;
+    for (Integer ip : as2)
+    {
+      mapseq2[i++] = ip;
+    }
+    ;
+    if (lastmatch)
+    {
+      mapseq1[mapseq1.length - 1] = alignpos;
+      mapseq2[mapseq2.length - 1] = pdbpos;
+    }
+    MapList map = new MapList(mapseq1, mapseq2, 1, 1);
+
+    jalview.datamodel.Mapping mapping = new Mapping(map);
+    mapping.setTo(s2);
+    return mapping;
+  }
+
+  /**
    * compute the PID vector used by the redundancy filter.
    * 
    * @param originalSequences