Merge branch 'features/r2_11_2_alphafold/JAL-629' into features/JAL-3858_PAEsInProjects

[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 af8dfc3..0349417 100644 (file)
--- a/src/jalview/ws/datamodel/alphafold/PAEContactMatrix.java
+++ b/src/jalview/ws/datamodel/alphafold/PAEContactMatrix.java
@@ -9,12 +9,18 @@ import jalview.datamodel.ContactListImpl;
 import jalview.datamodel.ContactListProviderI;
 import jalview.datamodel.ContactMatrixI;
 import jalview.datamodel.SequenceI;
+import jalview.util.MapUtils;
 
 public class PAEContactMatrix implements ContactMatrixI
 {
 
   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;
@@ -23,18 +29,102 @@ public class PAEContactMatrix implements ContactMatrixI
 
   float maxscore;
 
-  @SuppressWarnings("unchecked")
-  public PAEContactMatrix(SequenceI _refSeq, Map<String, Object> pae_obj)
-          throws Exception
+  private void setRefSeq(SequenceI _refSeq)
   {
     refSeq = _refSeq;
     while (refSeq.getDatasetSequence() != null)
     {
       refSeq = refSeq.getDatasetSequence();
     }
+    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
-    int length = _refSeq.getEnd() - _refSeq.getStart() + 1;
+    
+    if (!MapUtils.containsAKey(pae_obj, "predicted_aligned_error", "pae"))
+    {
+      parse_version_1_pAE(pae_obj);
+      return;
+    }
+    else
+    {
+      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
+            .getFirst(pae_obj, "predicted_aligned_error", "pae"))
+            .iterator();
+    int row = 0, col = 0;
+    while (scoreRows.hasNext())
+    {
+      Iterator<Long> scores = scoreRows.next().iterator();
+      while (scores.hasNext())
+      {
+        Object d = scores.next();
+        if (d instanceof Double)
+          elements[row][col++] = ((Double) d).longValue();
+        else
+          elements[row][col++] = (float) d;
+      }
+      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
+   */
+  @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
     // dataset refSeq
     Iterator<Long> rows = ((List<Long>) pae_obj.get("residue1")).iterator();
@@ -59,8 +149,8 @@ public class PAEContactMatrix implements ContactMatrixI
       elements[row - 1][col - 1] = escore;
     }
 
-    maxscore = ((Double) pae_obj.get("max_predicted_aligned_error"))
-            .floatValue();
+    maxscore = ((Double) MapUtils.getFirst(pae_obj,
+            "max_predicted_aligned_error", "max_pae")).floatValue();
   }
 
   @Override
@@ -74,9 +164,15 @@ public class PAEContactMatrix implements ContactMatrixI
     return new ContactListImpl(new ContactListProviderI()
     {
       @Override
+      public int getPosition()
+      {
+        return _column;
+      }
+
+      @Override
       public int getContactHeight()
       {
-        return maxcol - 1;
+        return maxcol-1;
       }
 
       @Override
@@ -86,7 +182,6 @@ public class PAEContactMatrix implements ContactMatrixI
         {
           return -1;
         }
-        // TODO Auto-generated method stub
         return elements[_column][column];
       }
     });
@@ -116,4 +211,32 @@ 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";
+  }
+
+  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;
+  }
 }