JAL-4134 allow tree groups to be stored/recovered on contact matrix for groupwise...

[jalview.git] / src / jalview / ws / datamodel / alphafold / PAEContactMatrix.java

diff --git a/src/jalview/ws/datamodel/alphafold/PAEContactMatrix.java b/src/jalview/ws/datamodel/alphafold/PAEContactMatrix.java
index 272d21d..30c77d2 100644 (file)
--- a/src/jalview/ws/datamodel/alphafold/PAEContactMatrix.java
+++ b/src/jalview/ws/datamodel/alphafold/PAEContactMatrix.java
@@ -1,9 +1,14 @@
 package jalview.ws.datamodel.alphafold;
 
+import java.util.ArrayList;
+import java.util.BitSet;
 import java.util.Iterator;
 import java.util.List;
 import java.util.Map;
 
+import jalview.analysis.AverageDistanceEngine;
+import jalview.bin.Console;
+import jalview.datamodel.BinaryNode;
 import jalview.datamodel.ContactListI;
 import jalview.datamodel.ContactListImpl;
 import jalview.datamodel.ContactListProviderI;
@@ -29,17 +34,21 @@ public class PAEContactMatrix implements ContactMatrixI
 
   float maxscore;
 
-  @SuppressWarnings("unchecked")
-  public PAEContactMatrix(SequenceI _refSeq, Map<String, Object> pae_obj)
+  private void setRefSeq(SequenceI _refSeq)
   {
     refSeq = _refSeq;
     while (refSeq.getDatasetSequence() != null)
     {
       refSeq = refSeq.getDatasetSequence();
     }
-    // convert the lists to primitive arrays and store
     length = _refSeq.getEnd() - _refSeq.getStart() + 1;
-
+  }
+  @SuppressWarnings("unchecked")
+  public PAEContactMatrix(SequenceI _refSeq, Map<String, Object> pae_obj)
+  {
+    setRefSeq(_refSeq);
+    // convert the lists to primitive arrays and store
+    
     if (!MapUtils.containsAKey(pae_obj, "predicted_aligned_error", "pae"))
     {
       parse_version_1_pAE(pae_obj);
@@ -50,28 +59,60 @@ public class PAEContactMatrix implements ContactMatrixI
       parse_version_2_pAE(pae_obj);
     }
   }
+  /**
+   * construct a sequence associated PAE matrix directly from a float array
+   * @param _refSeq
+   * @param matrix
+   */
+  public PAEContactMatrix(SequenceI _refSeq, float[][] matrix)
+  {
+    setRefSeq(_refSeq);
+    maxcol=0;
+    for (float[] row:matrix)
+    {
+      if (row.length>maxcol)
+      {
+        maxcol=row.length;
+      }
+      maxscore=row[0];
+      for (float f:row)
+      {
+        if (maxscore<f) {
+          maxscore=f;
+        }
+      }
+    }
+    maxrow=matrix.length;
+    elements = matrix;
+    
+  }
 
   /**
    * parse a sane JSON representation of the pAE
    * 
    * @param pae_obj
    */
+  @SuppressWarnings("unchecked")
   private void parse_version_2_pAE(Map<String, Object> pae_obj)
   {
-    elements = new float[length][length];
     // this is never going to be reached by the integer rounding.. or is it ?
     maxscore = ((Double) MapUtils.getFirst(pae_obj,
             "max_predicted_aligned_error", "max_pae")).floatValue();
-    Iterator<List<Long>> scoreRows = ((List<List<Long>>) MapUtils
+    List<List<Long>> scoreRows = ((List<List<Long>>) MapUtils
             .getFirst(pae_obj, "predicted_aligned_error", "pae"))
-            .iterator();
+            ;
+    elements = new float[scoreRows.size()][scoreRows.size()];
     int row = 0, col = 0;
-    while (scoreRows.hasNext())
+    for (List<Long> scoreRow:scoreRows)
     {
-      Iterator<Long> scores = scoreRows.next().iterator();
+      Iterator<Long> scores = scoreRow.iterator();
       while (scores.hasNext())
       {
-        elements[row][col++] = scores.next();
+        Object d = scores.next();
+        if (d instanceof Double)
+          elements[row][col++] = ((Double) d).longValue();
+        else
+          elements[row][col++] = (float) ((Long)d).longValue();
       }
       row++;
       col = 0;
@@ -86,6 +127,7 @@ public class PAEContactMatrix implements ContactMatrixI
    * 
    * @param pae_obj
    */
+  @SuppressWarnings("unchecked")
   private void parse_version_1_pAE(Map<String, Object> pae_obj)
   {
     // assume indices are with respect to range defined by _refSeq on the
@@ -94,7 +136,7 @@ public class PAEContactMatrix implements ContactMatrixI
     Iterator<Long> cols = ((List<Long>) pae_obj.get("residue2")).iterator();
     Iterator<Double> scores = ((List<Double>) pae_obj.get("distance"))
             .iterator();
-
+    // assume square matrix
     elements = new float[length][length];
     while (scores.hasNext())
     {
@@ -135,7 +177,7 @@ public class PAEContactMatrix implements ContactMatrixI
       @Override
       public int getContactHeight()
       {
-        return maxcol - 1;
+        return maxcol-1;
       }
 
       @Override
@@ -145,7 +187,6 @@ public class PAEContactMatrix implements ContactMatrixI
         {
           return -1;
         }
-        // TODO Auto-generated method stub
         return elements[_column][column];
       }
     });
@@ -186,4 +227,77 @@ public class PAEContactMatrix implements ContactMatrixI
   {
     return "pAE Matrix";
   }
+
+  public static final String PAEMATRIX="PAE_MATRIX";
+  @Override
+  public String getType()
+  {
+    return PAEMATRIX;
+  }
+  @Override
+  public int getWidth()
+  {
+    return length;
+  }
+  @Override
+  public int getHeight()
+  {
+    return length;
+  }
+  List<BitSet> groups=null;
+  @Override
+  public boolean hasGroups()
+  {
+    return groups!=null;
+  }
+  String newick=null;
+  public String getNewickString()
+  {
+    return newick;
+  }
+  public void makeGroups(float thresh,boolean abs)
+  {
+    AverageDistanceEngine clusterer = new AverageDistanceEngine(null, null, this);
+    double height = clusterer.findHeight(clusterer.getTopNode());
+    newick = new jalview.io.NewickFile(clusterer.getTopNode(),false,true).print();
+
+    Console.trace("Newick string\n"+newick);
+
+    List<BinaryNode> nodegroups;
+    if (abs ? height > thresh : 0 < thresh && thresh < 1)
+    {
+      float cut = abs ? (float) (thresh / height) : thresh;
+      Console.debug("Threshold "+cut+" for height="+height);
+
+      nodegroups = clusterer.groupNodes(cut);
+    }
+    else
+    {
+      nodegroups = new ArrayList<BinaryNode>();
+      nodegroups.add(clusterer.getTopNode());
+    }
+
+    groups = new ArrayList<>();
+    for (BinaryNode root:nodegroups)
+    {
+      BitSet gpset=new BitSet();
+      for (BinaryNode leaf:clusterer.findLeaves(root))
+      {
+        gpset.set((Integer)leaf.element());
+      }
+      groups.add(gpset);
+    }
+  }
+  
+  @Override
+  public BitSet getGroupsFor(int column)
+  {
+    for (BitSet gp:groups) {
+      if (gp.get(column))
+      {
+        return gp;
+      }
+    }
+    return ContactMatrixI.super.getGroupsFor(column);
+  }
 }