JAL-1827 refactored to handle missing residues better

[jalview.git] / test / jalview / ext / jmol / PDBFileWithJmolTest.java

diff --git a/test/jalview/ext/jmol/PDBFileWithJmolTest.java b/test/jalview/ext/jmol/PDBFileWithJmolTest.java
index 4ccb6e3..fd02d00 100644 (file)
--- a/test/jalview/ext/jmol/PDBFileWithJmolTest.java
+++ b/test/jalview/ext/jmol/PDBFileWithJmolTest.java
@@ -44,11 +44,40 @@ import MCview.PDBfile;
  */
 public class PDBFileWithJmolTest
 {
+  /*
+   * 1GAQ has been reduced to alpha carbons only
+   * 1QCF is the full PDB file including headers, HETATM etc
+   */
   String[] testFile = new String[] { "./examples/1GAQ.txt",
       "./test/jalview/ext/jmol/1QCF.pdb" }; // ,
 
-  // "./examples/DNMT1_MOUSE.pdb"
-  // };
+  //@formatter:off
+  // a modified and very cut-down extract of 4UJ4
+  String pdbWithChainBreak =
+     "HEADER    TRANSPORT PROTEIN                       08-APR-15   4UJ4\n" +
+     // chain B has missing residues; these should all go in the same sequence:
+     "ATOM   1909  CA  VAL B 358      21.329 -19.739 -67.740  1.00201.05           C\n" +
+     "ATOM   1916  CA  GLY B 359      21.694 -23.563 -67.661  1.00198.09           C\n" +
+     "ATOM   1920  CA  LYS B 367      32.471 -12.135 -77.100  1.00257.97           C\n" +
+     "ATOM   1925  CA  ALA B 368      31.032  -9.324 -74.946  1.00276.01           C\n" +
+     // switch to chain C; should be a separate sequence
+     "ATOM   1930  CA  SER C 369      32.589  -7.517 -71.978  1.00265.44           C\n" +
+     "ATOM   1936  CA  ALA C 370      31.650  -6.849 -68.346  1.00249.48           C\n";
+  //@formatter:on
+
+  //@formatter:off
+  // a very cut-down extract of 1ejg
+  String pdbWithAltLoc =
+     "HEADER    TRANSPORT PROTEIN                       08-APR-15   1EJG\n" +
+     "ATOM    448  CA  ALA A  24       6.619  16.195   1.970  1.00  1.65           C\n" +
+     "ATOM    458  CA ALEU A  25       3.048  14.822   1.781  0.57  1.48           C\n" +
+     // alternative residue 25 entries (with ILE instead of LEU) should be ignored:
+     "ATOM    478  CA BILE A  25       3.048  14.822   1.781  0.21  1.48           C\n" +
+     // including the next altloc causes the unit test to fail but it works with the full file
+     // not sure why!
+     //     "ATOM    479  CA CILE A  25       3.048  14.822   1.781  0.22  1.48           C\n" +
+     "ATOM    512  CA  CYS A  26       4.137  11.461   3.154  1.00  1.52           C\n";
+  //@formatter:on
 
   @BeforeMethod(alwaysRun = true)
   public void setUp()
@@ -133,4 +162,48 @@ public class PDBFileWithJmolTest
             "Secondary structure not associated for sequence "
                     + sq.getName(), sq.getAnnotation()[0].sequenceRef == sq);
   }
+
+  /**
+   * Test parsing a chain with missing residues
+   * 
+   * @throws Exception
+   */
+  @Test(groups = { "Functional" })
+  public void testParse_missingResidues() throws Exception
+  {
+    PDBfile mctest = new PDBfile(false, false, false, pdbWithChainBreak,
+            AppletFormatAdapter.PASTE);
+    PDBFileWithJmol jtest = new PDBFileWithJmol(pdbWithChainBreak,
+            jalview.io.AppletFormatAdapter.PASTE);
+    Vector<SequenceI> seqs = jtest.getSeqs();
+    Vector<SequenceI> mcseqs = mctest.getSeqs();
+
+    assertEquals("Failed to find 2 sequences\n", 2, seqs.size());
+    assertEquals("Failed to find 2 sequences\n", 2, mcseqs.size());
+    assertEquals("VGKA", seqs.get(0).getSequenceAsString());
+    assertEquals("VGKA", mcseqs.get(0).getSequenceAsString());
+    assertEquals("SA", seqs.get(1).getSequenceAsString());
+    assertEquals("SA", mcseqs.get(1).getSequenceAsString());
+  }
+
+  /**
+   * Test parsing a chain with 'altloc' residues
+   * 
+   * @throws Exception
+   */
+  @Test(groups = { "Functional" })
+  public void testParse_alternativeResidues() throws Exception
+  {
+    PDBfile mctest = new PDBfile(false, false, false, pdbWithAltLoc,
+            AppletFormatAdapter.PASTE);
+    PDBFileWithJmol jtest = new PDBFileWithJmol(pdbWithAltLoc,
+            jalview.io.AppletFormatAdapter.PASTE);
+    Vector<SequenceI> seqs = jtest.getSeqs();
+    Vector<SequenceI> mcseqs = mctest.getSeqs();
+  
+    assertEquals("Failed to find 1 sequence\n", 1, seqs.size());
+    assertEquals("Failed to find 1 sequence\n", 1, mcseqs.size());
+    assertEquals("ALC", seqs.get(0).getSequenceAsString());
+    assertEquals("ALC", mcseqs.get(0).getSequenceAsString());
+  }
 }