JAL-3855 pAE v2 import
authorJim Procter <j.procter@dundee.ac.uk>
Sat, 15 Oct 2022 10:09:52 +0000 (11:09 +0100)
committerJim Procter <j.procter@dundee.ac.uk>
Sat, 15 Oct 2022 10:09:52 +0000 (11:09 +0100)
src/jalview/ws/datamodel/alphafold/PAEContactMatrix.java

index af8dfc3..8b6771e 100644 (file)
@@ -15,6 +15,11 @@ 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;
@@ -33,8 +38,53 @@ public class PAEContactMatrix implements ContactMatrixI
       refSeq = refSeq.getDatasetSequence();
     }
     // convert the lists to primitive arrays and store
-    int length = _refSeq.getEnd() - _refSeq.getStart() + 1;
+    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())
+      {
+        elements[row][col++] = scores.next();
+      }
+      row++;
+      col = 0;
+    }
+  }
 
+  /**
+   * 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();