X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;ds=sidebyside;f=src%2Fjalview%2Fws%2Fdatamodel%2Falphafold%2FPAEContactMatrix.java;h=dcd20225e83e38efd36f06dfda61b4d88412cd5f;hb=cd669a0e8c7b91b379bca8fe6e702cf0fcbd1ce0;hp=48071bdb8f8a1ea7f8dfcf4757403f3cb2f88aca;hpb=7f98aeaa190d77fd40dcf80d55faedcace024d6f;p=jalview.git

diff --git a/src/jalview/ws/datamodel/alphafold/PAEContactMatrix.java b/src/jalview/ws/datamodel/alphafold/PAEContactMatrix.java
index 48071bd..dcd2022 100644
--- a/src/jalview/ws/datamodel/alphafold/PAEContactMatrix.java
+++ b/src/jalview/ws/datamodel/alphafold/PAEContactMatrix.java
@@ -1,49 +1,54 @@
 package jalview.ws.datamodel.alphafold;
 
+import java.io.File;
+import java.io.FileInputStream;
+import java.io.IOException;
 import java.util.Iterator;
 import java.util.List;
 import java.util.Map;
 
+import org.json.simple.JSONObject;
+
 import jalview.datamodel.ContactListI;
 import jalview.datamodel.ContactListImpl;
 import jalview.datamodel.ContactListProviderI;
 import jalview.datamodel.ContactMatrixI;
+import jalview.datamodel.FloatContactMatrix;
+import jalview.datamodel.GroupSet;
+import jalview.datamodel.SequenceDummy;
 import jalview.datamodel.SequenceI;
+import jalview.io.FileFormatException;
+import jalview.util.MapList;
 import jalview.util.MapUtils;
+import jalview.ws.dbsources.EBIAlfaFold;
 
-public class PAEContactMatrix implements ContactMatrixI
+/**
+ * routines and class for holding predicted alignment error matrices as produced
+ * by alphafold et al.
+ * 
+ * getContactList(column) returns the vector of predicted alignment errors for
+ * reference position given by column getElementAt(column, i) returns the
+ * predicted superposition error for the ith position when column is used as
+ * reference
+ * 
+ * Many thanks to Ora Schueler Furman for noticing that earlier development
+ * versions did not show the PAE oriented correctly
+ *
+ * @author jprocter
+ *
+ */
+public class PAEContactMatrix extends
+        MappableContactMatrix<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;
-
-  float[][] elements;
 
-  float maxscore;
-
-  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)
+          throws FileFormatException
   {
     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);
@@ -54,66 +59,103 @@ 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)
   {
+    mappedMatrix=new FloatContactMatrix(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
+   * new matrix with specific mapping to a reference sequence
+   * 
+   * @param newRefSeq
+   * @param newFromMapList
+   * @param elements2
+   * @param grps2
+   */
+  public PAEContactMatrix(SequenceI newRefSeq, MapList newFromMapList,
+          float[][] elements2, GroupSet grps2)
+  {
+    this(newRefSeq, new FloatContactMatrix(elements2,grps2));
+    toSeq = newFromMapList;
+  }
+
+  public PAEContactMatrix(SequenceI _refSeq,
+          ContactMatrixI floatContactMatrix)
+  {
+    mappedMatrix = floatContactMatrix;
+    setRefSeq(_refSeq);
+  }
+  public PAEContactMatrix(SequenceI _refSeq, MapList newFromMapList,
+          ContactMatrixI floatContactMatrix)
+  {
+    mappedMatrix = floatContactMatrix;
+    setRefSeq(_refSeq);
+    toSeq = newFromMapList;
+  }
+  
+  @Override
+  protected PAEContactMatrix newMappableContactMatrix(SequenceI newRefSeq,
+          MapList newFromMapList)
+  {
+    return new PAEContactMatrix(newRefSeq, newFromMapList, mappedMatrix);
+  }
+
+  /**
+   * parse a sane JSON representation of the pAE and update the mappedMatrix
    * 
    * @param pae_obj
    */
   @SuppressWarnings("unchecked")
   private void parse_version_2_pAE(Map<String, Object> pae_obj)
   {
-    // 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();
+    float maxscore = -1;
+    // look for a maxscore element - if there is one...
+    try
+    {
+      // 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();
+    } catch (Throwable t)
+    {
+      // ignore if a key is not found.
+    }
     List<List<Long>> scoreRows = ((List<List<Long>>) MapUtils
-            .getFirst(pae_obj, "predicted_aligned_error", "pae"))
-            ;
-    elements = new float[scoreRows.size()][scoreRows.size()];
+            .getFirst(pae_obj, "predicted_aligned_error", "pae"));
+    float[][] elements = new float[scoreRows.size()][scoreRows.size()];
     int row = 0, col = 0;
-    for (List<Long> scoreRow:scoreRows)
+    for (List<Long> scoreRow : scoreRows)
     {
       Iterator<Long> scores = scoreRow.iterator();
       while (scores.hasNext())
       {
         Object d = scores.next();
         if (d instanceof Double)
-          elements[row][col++] = ((Double) d).longValue();
+        {
+          elements[col][row] = ((Double) d).longValue();
+        }
         else
-          elements[row][col++] = (float) ((Long)d).longValue();
+        {
+          elements[col][row] = (float) ((Long) d).longValue();
+        }
+
+        if (maxscore < elements[col][row])
+        {
+          maxscore = elements[col][row];
+        }
+        col++;
       }
       row++;
       col = 0;
     }
-    maxcol = length;
-    maxrow = length;
+    mappedMatrix=new FloatContactMatrix(elements);
   }
 
   /**
@@ -129,13 +171,11 @@ public class PAEContactMatrix implements ContactMatrixI
     // 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();
-    // assume square matrix
-    elements = new float[length][length];
-    while (scores.hasNext())
+    // two pass - to allocate the elements array
+    
+    int maxrow=-1,maxcol=-1;
+    while (rows.hasNext())
     {
-      float escore = scores.next().floatValue();
       int row = rows.next().intValue();
       int col = cols.next().intValue();
       if (maxrow < row)
@@ -146,97 +186,96 @@ public class PAEContactMatrix implements ContactMatrixI
       {
         maxcol = col;
       }
-      elements[row - 1][col - 1] = escore;
-    }
 
-    maxscore = ((Double) MapUtils.getFirst(pae_obj,
-            "max_predicted_aligned_error", "max_pae")).floatValue();
-  }
-
-  @Override
-  public ContactListI getContactList(final int _column)
-  {
-    if (_column < 0 || _column >= elements.length)
-    {
-      return null;
     }
-
-    return new ContactListImpl(new ContactListProviderI()
+    rows = ((List<Long>) pae_obj.get("residue1")).iterator();
+    cols = ((List<Long>) pae_obj.get("residue2")).iterator();
+    Iterator<Double> scores = ((List<Double>) pae_obj.get("distance"))
+            .iterator();
+    float[][] elements = new float[maxcol][maxrow];
+    while (scores.hasNext())
     {
-      @Override
-      public int getPosition()
+      float escore = scores.next().floatValue();
+      int row = rows.next().intValue();
+      int col = cols.next().intValue();
+      if (maxrow < row)
       {
-        return _column;
+        maxrow = row;
       }
-
-      @Override
-      public int getContactHeight()
+      if (maxcol < col)
       {
-        return maxcol-1;
+        maxcol = col;
       }
+      elements[col - 1][row-1] = escore;
+    }
 
-      @Override
-      public double getContactAt(int column)
-      {
-        if (column < 0 || column >= elements[_column].length)
-        {
-          return -1;
-        }
-        return elements[_column][column];
-      }
-    });
+    mappedMatrix=new FloatContactMatrix(elements);
   }
 
   @Override
-  public float getMin()
+  public String getAnnotDescr()
   {
-    return 0;
+    return "Predicted Alignment Error"
+            + ((refSeq == null) ? "" : (" for " + refSeq.getName()));
   }
 
   @Override
-  public float getMax()
+  public String getAnnotLabel()
   {
-    return maxscore;
+    StringBuilder label = new StringBuilder("PAE Matrix");
+    // if (this.getReferenceSeq() != null)
+    // {
+    // label.append(":").append(this.getReferenceSeq().getDisplayId(false));
+    // }
+    return label.toString();
   }
 
-  @Override
-  public boolean hasReferenceSeq()
-  {
-    return (refSeq != null);
-  }
+  public static final String PAEMATRIX = "PAE_MATRIX";
 
   @Override
-  public SequenceI getReferenceSeq()
+  public String getType()
   {
-    return refSeq;
+    return PAEMATRIX;
   }
 
-  @Override
-  public String getAnnotDescr()
-  {
-    return "Predicted Alignment Error for " + refSeq.getName();
-  }
 
-  @Override
-  public String getAnnotLabel()
+  public static void validateContactMatrixFile(String fileName)
+          throws FileFormatException, IOException
   {
-    return "pAE Matrix";
-  }
+    FileInputStream infile = null;
+    try
+    {
+      infile = new FileInputStream(new File(fileName));
+    } catch (Throwable t)
+    {
+      new IOException("Couldn't open " + fileName, t);
+    }
+    JSONObject paeDict = null;
+    try
+    {
+      paeDict = EBIAlfaFold.parseJSONtoPAEContactMatrix(infile);
+    } catch (Throwable t)
+    {
+      new FileFormatException("Couldn't parse " + fileName
+              + " as a JSON dict or array containing a dict");
+    }
 
-  public static final String PAEMATRIX="PAE_MATRIX";
-  @Override
-  public String getType()
-  {
-    return PAEMATRIX;
+    PAEContactMatrix matrix = new PAEContactMatrix(
+            new SequenceDummy("Predicted"), (Map<String, Object>) paeDict);
+    if (matrix.getWidth() <= 0)
+    {
+      throw new FileFormatException(
+              "No data in PAE matrix read from '" + fileName + "'");
+    }
   }
   @Override
-  public int getWidth()
+  public boolean equals(Object obj)
   {
-    return length;
+    return super.equals(obj);
   }
   @Override
-  public int getHeight()
+  public int hashCode()
   {
-    return length;
+    return super.hashCode();
   }
 }