JAL-629 implementation of --tempfac options

[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 e61af44..3298a96 100644 (file)
--- a/src/jalview/ws/datamodel/alphafold/PAEContactMatrix.java
+++ b/src/jalview/ws/datamodel/alphafold/PAEContactMatrix.java
@@ -8,69 +8,138 @@ import jalview.datamodel.ContactListI;
 import jalview.datamodel.ContactListImpl;
 import jalview.datamodel.ContactListProviderI;
 import jalview.datamodel.ContactMatrixI;
-import jalview.datamodel.ContactRange;
 import jalview.datamodel.SequenceI;
 
 public class PAEContactMatrix implements ContactMatrixI
 {
 
-  SequenceI refSeq=null;
-  int maxrow=0,maxcol=0;
-  int[] indices1,indices2;
+  SequenceI refSeq = null;
+
+  /**
+   * the length that refSeq is expected to be (excluding gaps, of course)
+   */
+  int length;
+
+  int maxrow = 0, maxcol = 0;
+
+  int[] indices1, indices2;
+
   float[][] elements;
+
   float maxscore;
-  
+
   @SuppressWarnings("unchecked")
-  public PAEContactMatrix(SequenceI _refSeq, Map<String, Object> pae_obj) throws Exception
+  public PAEContactMatrix(SequenceI _refSeq, Map<String, Object> pae_obj)
   {
     refSeq = _refSeq;
-    while (refSeq.getDatasetSequence()!=null)
+    while (refSeq.getDatasetSequence() != null)
     {
-      refSeq=refSeq.getDatasetSequence();
+      refSeq = refSeq.getDatasetSequence();
     }
     // convert the lists to primitive arrays and store
-    int length = _refSeq.getEnd()-_refSeq.getStart()+1;
-    
-    // assume indices are with respect to range defined by _refSeq on the dataset refSeq
-    Iterator<Long> rows = ((List<Long>)pae_obj.get("residue1")).iterator();
-    Iterator<Long> cols = ((List<Long>)pae_obj.get("residue2")).iterator();
-    Iterator<Double> scores = ((List<Double>)pae_obj.get("distance")).iterator();
-    
-    elements=new float[length][length];
-    while (scores.hasNext()) {
-      float escore=scores.next().floatValue();
-      int row=rows.next().intValue();
-      int col=cols.next().intValue();
-      if (maxrow<row)
+    length = _refSeq.getEnd() - _refSeq.getStart() + 1;
+
+    if (!pae_obj.containsKey("predicted_aligned_error"))
+    {
+      parse_version_1_pAE(pae_obj);
+      return;
+    }
+    else
+    {
+      parse_version_2_pAE(pae_obj);
+    }
+  }
+
+  /**
+   * parse a sane JSON representation of the pAE
+   * 
+   * @param pae_obj
+   */
+  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) pae_obj.get("max_predicted_aligned_error"))
+            .floatValue();
+    Iterator<List<Long>> scoreRows = ((List<List<Long>>) pae_obj
+            .get("predicted_aligned_error")).iterator();
+    int row = 0, col = 0;
+    while (scoreRows.hasNext())
+    {
+      Iterator<Long> scores = scoreRows.next().iterator();
+      while (scores.hasNext())
       {
-        maxrow=row;
+        elements[row][col++] = scores.next();
       }
-      if (maxcol<col)
+      row++;
+      col = 0;
+    }
+    maxcol = length;
+    maxrow = length;
+  }
+
+  /**
+   * v1 format got ditched 28th July 2022 see
+   * https://alphafold.ebi.ac.uk/faq#:~:text=We%20updated%20the%20PAE%20JSON%20file%20format%20on%2028th%20July%202022
+   * 
+   * @param pae_obj
+   */
+  private void parse_version_1_pAE(Map<String, Object> pae_obj)
+  {
+    // assume indices are with respect to range defined by _refSeq on the
+    // dataset refSeq
+    Iterator<Long> rows = ((List<Long>) pae_obj.get("residue1")).iterator();
+    Iterator<Long> cols = ((List<Long>) pae_obj.get("residue2")).iterator();
+    Iterator<Double> scores = ((List<Double>) pae_obj.get("distance"))
+            .iterator();
+
+    elements = new float[length][length];
+    while (scores.hasNext())
+    {
+      float escore = scores.next().floatValue();
+      int row = rows.next().intValue();
+      int col = cols.next().intValue();
+      if (maxrow < row)
+      {
+        maxrow = row;
+      }
+      if (maxcol < col)
       {
-        maxcol=col;
+        maxcol = col;
       }
-      elements[row-1][col-1]=escore;
+      elements[row - 1][col - 1] = escore;
     }
-    
-    maxscore = ((Double) pae_obj.get("max_predicted_aligned_error")).floatValue();
+
+    maxscore = ((Double) pae_obj.get("max_predicted_aligned_error"))
+            .floatValue();
   }
 
   @Override
   public ContactListI getContactList(final int _column)
   {
-    
-    return new ContactListImpl(new ContactListProviderI() 
+    if (_column < 0 || _column >= elements.length)
+    {
+      return null;
+    }
+
+    return new ContactListImpl(new ContactListProviderI()
     {
       @Override
+      public int getPosition()
+      {
+        return _column;
+      }
+
+      @Override
       public int getContactHeight()
       {
-        return maxcol-1;
+        return maxcol - 1;
       }
-      
+
       @Override
       public double getContactAt(int column)
       {
-        if (column<0 || column>=elements[_column].length)
+        if (column < 0 || column >= elements[_column].length)
         {
           return -1;
         }
@@ -95,7 +164,7 @@ public class PAEContactMatrix implements ContactMatrixI
   @Override
   public boolean hasReferenceSeq()
   {
-    return (refSeq!=null);
+    return (refSeq != null);
   }
 
   @Override
@@ -104,4 +173,15 @@ public class PAEContactMatrix implements ContactMatrixI
     return refSeq;
   }
 
+  @Override
+  public String getAnnotDescr()
+  {
+    return "Predicted Alignment Error for " + refSeq.getName();
+  }
+
+  @Override
+  public String getAnnotLabel()
+  {
+    return "pAE Matrix";
+  }
 }