added default ClustalWS or MuscleWS to remove gaps.

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

diff --git a/src/jalview/analysis/NJTree.java b/src/jalview/analysis/NJTree.java
index 4aeee3d..d56464c 100755 (executable)
--- a/src/jalview/analysis/NJTree.java
+++ b/src/jalview/analysis/NJTree.java
@@ -5,6 +5,8 @@ import jalview.util.*;
 import jalview.schemes.ResidueProperties;\r
 import java.util.*;\r
 \r
+import jalview.io.NewickFile;\r
+\r
 public class NJTree {\r
 \r
   Vector cluster;\r
@@ -46,6 +48,79 @@ public class NJTree {
     maxheight = findHeight(top);\r
   }\r
 \r
+  public String toString()\r
+  {\r
+    jalview.io.NewickFile fout = new jalview.io.NewickFile(getTopNode());\r
+    return fout.print(false,true); // distances only\r
+  }\r
+\r
+  public NJTree(SequenceI[] seqs, NewickFile treefile) {\r
+    top = treefile.getTree();\r
+    maxheight = findHeight(top);\r
+    SequenceIdMatcher algnIds = new SequenceIdMatcher(seqs);\r
+\r
+    Vector leaves = new Vector();\r
+    findLeaves(top, leaves);\r
+\r
+    int i = 0;\r
+    int namesleft = seqs.length;\r
+\r
+    SequenceNode j;\r
+    SequenceI nam;\r
+    String realnam;\r
+    while (i < leaves.size())\r
+    {\r
+      j = (SequenceNode) leaves.elementAt(i++);\r
+      realnam = j.getName();\r
+      nam = null;\r
+      if (namesleft>-1)\r
+        nam = algnIds.findIdMatch(realnam);\r
+      if (nam != null) {\r
+        j.setElement(nam);\r
+        namesleft--;\r
+      } else {\r
+        j.setElement(new Sequence(realnam, "THISISAPLACEHLDER"));\r
+        j.setPlaceholder(true);\r
+\r
+      }\r
+    }\r
+  }\r
+\r
+  /**\r
+   *\r
+   * used when the alignment associated to a tree has changed.\r
+   *\r
+   * @param alignment Vector\r
+   */\r
+  public void UpdatePlaceHolders(Vector alignment) {\r
+    Vector leaves = new Vector();\r
+    findLeaves(top, leaves);\r
+    int sz = leaves.size();\r
+    SequenceIdMatcher seqmatcher=null;\r
+    int i=0;\r
+    while (i<sz) {\r
+      SequenceNode leaf = (SequenceNode) leaves.elementAt(i++);\r
+      if (alignment.contains(leaf.element()))\r
+        leaf.setPlaceholder(false);\r
+      else {\r
+        if (seqmatcher==null) {\r
+          // Only create this the first time we need it\r
+          SequenceI[] seqs = new SequenceI[alignment.size()];\r
+          for (int j=0; j<seqs.length; j++)\r
+            seqs[j] = (SequenceI) alignment.elementAt(j);\r
+          seqmatcher = new SequenceIdMatcher(seqs);\r
+        }\r
+        SequenceI nam = seqmatcher.findIdMatch(leaf.getName());\r
+        if (nam!=null) {\r
+          leaf.setPlaceholder(false);\r
+          leaf.setElement(nam);\r
+        } else {\r
+          leaf.setPlaceholder(true);\r
+        }\r
+      }\r
+    }\r
+  }\r
+\r
   public NJTree(SequenceI[] sequence,int start, int end) {\r
     this(sequence,"NJ","BL",start,end);\r
   }\r
@@ -331,14 +406,13 @@ public class NJTree {
   public float[][] findDistances() {\r
 \r
     float[][] distance = new float[noseqs][noseqs];\r
-\r
     if (pwtype.equals("PID")) {\r
       for (int i = 0; i < noseqs-1; i++) {\r
         for (int j = i; j < noseqs; j++) {\r
           if (j==i) {\r
             distance[i][i] = 0;\r
           } else {\r
-            distance[i][j] = 100-Comparison.compare(sequence[i],sequence[j],start,end);\r
+            distance[i][j] = 100-Comparison.PID(sequence[i], sequence[j]);\r
             distance[j][i] = distance[i][j];\r
           }\r
         }\r
@@ -356,7 +430,7 @@ public class NJTree {
                   k + 1),\r
                                                 sequence[j].getSequence(k,\r
                   k + 1));\r
-            }catch(Exception ex){System.out.println("err creating BLOSUM62 tree");}\r
+            }catch(Exception ex){System.err.println("err creating BLOSUM62 tree");ex.printStackTrace();}\r
           }\r
           distance[i][j] = (float)score;\r
           if (score > maxscore) {\r
@@ -448,6 +522,11 @@ public class NJTree {
     return found;\r
   }\r
 \r
+  /**\r
+   * printNode is mainly for debugging purposes.\r
+   *\r
+   * @param node SequenceNode\r
+   */\r
   public void printNode(SequenceNode node) {\r
     if (node == null) {\r
       return;\r
@@ -470,8 +549,8 @@ public class NJTree {
 \r
       float dist = ((SequenceNode)node).dist;\r
       if (dist > maxDistValue) {\r
-         maxdist      = (SequenceNode)node;\r
-         maxDistValue = dist;\r
+          maxdist      = (SequenceNode)node;\r
+          maxDistValue = dist;\r
       }\r
     } else {\r
       findMaxDist((SequenceNode)node.left());\r
@@ -479,10 +558,10 @@ public class NJTree {
     }\r
   }\r
     public Vector getGroups() {\r
-       return groups;\r
+        return groups;\r
     }\r
     public float getMaxHeight() {\r
-       return maxheight;\r
+        return maxheight;\r
     }\r
   public void  groupNodes(SequenceNode node, float threshold) {\r
     if (node == null) {\r
@@ -571,8 +650,8 @@ public class NJTree {
   }\r
 \r
     public void reCount(SequenceNode node) {\r
-       ycount = 0;\r
-       _reCount(node);\r
+        ycount = 0;\r
+        _reCount(node);\r
     }\r
   public void _reCount(SequenceNode node) {\r
     if (node == null) {\r
@@ -596,13 +675,13 @@ public class NJTree {
 \r
   }\r
     public void swapNodes(SequenceNode node) {\r
-       if (node == null) {\r
-           return;\r
-       }\r
-       SequenceNode tmp = (SequenceNode)node.left();\r
+        if (node == null) {\r
+            return;\r
+        }\r
+        SequenceNode tmp = (SequenceNode)node.left();\r
 \r
-       node.setLeft(node.right());\r
-       node.setRight(tmp);\r
+        node.setLeft(node.right());\r
+        node.setRight(tmp);\r
     }\r
   public void changeDirection(SequenceNode node, SequenceNode dir) {\r
     if (node == null) {\r
@@ -642,13 +721,13 @@ public class NJTree {
     }\r
   }\r
     public void setMaxDist(SequenceNode node) {\r
-       this.maxdist = maxdist;\r
+        this.maxdist = maxdist;\r
     }\r
     public SequenceNode getMaxDist() {\r
-       return maxdist;\r
+        return maxdist;\r
     }\r
     public SequenceNode getTopNode() {\r
-       return top;\r
+        return top;\r
     }\r
 \r
 }\r