test/jalview/datamodel/xdb/embl/EmblFileTest.java

   1 package jalview.datamodel.xdb.embl;
   2
   3 import static org.testng.AssertJUnit.assertEquals;
   4 import static org.testng.AssertJUnit.assertFalse;
   5 import static org.testng.AssertJUnit.assertNull;
   6 import static org.testng.AssertJUnit.assertTrue;
   7
   8 import jalview.datamodel.DBRefEntry;
   9
  10 import java.io.StringReader;
  11 import java.util.Vector;
  12
  13 import org.testng.annotations.Test;
  14
  15 public class EmblFileTest
  16 {
  17   // adapted from http://www.ebi.ac.uk/Tools/dbfetch/dbfetch/embl/x53828/emblxml
  18   private static final String TESTDATA =
  19           "<?xml version=\"1.0\" encoding=\"UTF-8\" ?>"
  20           + "<EMBL_Services>"
  21           + "<entry accession=\"X53828\" version=\"3\" lastUpdated=\"2005-04-18\" releaseCreated=\"25\" releaseLastUpdated=\"83\">"
  22           + "<description>Chicken LDH-A mRNA for lactate dehydrogenase A chain (EC 1.1.1.27)</description>"
  23           + "<keyword>L-lactate dehydrogenase</keyword><keyword>chutney</keyword>"
  24           + "<dbreference db=\"EuropePMC\" primary=\"PMC1460223\" secondary=\"9649548\" />"
  25           + "<feature name=\"CDS\"><dbreference db=\"GOA\" primary=\"P00340\" secondary=\"2.1\" /><dbreference db=\"InterPro\" primary=\"IPR001236\" />"
  26           + "<qualifier name=\"note\"><value>L-lactate dehydrogenase A-chain</value><value>pickle</value></qualifier>"
  27           + "<qualifier name=\"translation\"><value>MSLKDHLIHN</value><evidence>Keith</evidence></qualifier>"
  28           + "<location type=\"single\" complement=\"true\">"
  29               + "<locationElement type=\"range\" accession=\"X53828\" version=\"1\" complement=\"false\">"
  30           + "<basePosition type=\"simple\">60</basePosition><basePosition type=\"join\">1058</basePosition>"
  31               + "</locationElement></location></feature>"
  32           + "<sequence type=\"mRNA\" version=\"2\">GTGACG</sequence></entry></EMBL_Services>";
  33
  34   @Test
  35   public void testGetEmblFile()
  36   {
  37     Vector<EmblEntry> entries = EmblFile.getEmblFile(
  38             new StringReader(TESTDATA)).getEntries();
  39     assertEquals(1, entries.size());
  40     EmblEntry entry = entries.get(0);
  41
  42     assertEquals("X53828", entry.getAccession());
  43     assertEquals(
  44             "Chicken LDH-A mRNA for lactate dehydrogenase A chain (EC 1.1.1.27)",
  45             entry.getDesc());
  46     assertEquals("2005-04-18", entry.getLastUpdated());
  47
  48     /*
  49      * FIXME these assertions fail - values are null - why?? Adding or removing
  50      * attributes in the test XML modifies behaviour. eg. inserting an attribute
  51      * _before_ lastUpdated results in a null value in this field.
  52      */
  53     // assertEquals("25", entry.getRCreated());
  54     // assertEquals("83", entry.getRLastUpdated());
  55
  56     assertEquals(2, entry.getKeywords().size());
  57     assertEquals("L-lactate dehydrogenase", entry.getKeywords().get(0));
  58     assertEquals("chutney", entry.getKeywords().get(1));
  59
  60     /*
  61      * dbrefs
  62      */
  63     assertEquals(1, entry.getDbRefs().size());
  64     DBRefEntry dbref = entry.getDbRefs().get(0);
  65     assertEquals("EuropePMC", dbref.getSource());
  66     assertEquals("PMC1460223", dbref.getAccessionId());
  67     assertEquals("9649548", dbref.getVersion());
  68
  69     /*
  70      * sequence features
  71      */
  72     assertEquals(1, entry.getFeatures().size());
  73     EmblFeature ef = entry.getFeatures().get(0);
  74     assertEquals("CDS", ef.getName());
  75     assertEquals(2, ef.getDbRefs().size());
  76     dbref = ef.getDbRefs().get(0);
  77     assertEquals("GOA", dbref.getSource());
  78     assertEquals("P00340", dbref.getAccessionId());
  79     assertEquals("2.1", dbref.getVersion());
  80     dbref = ef.getDbRefs().get(1);
  81     assertEquals("InterPro", dbref.getSource());
  82     assertEquals("IPR001236", dbref.getAccessionId());
  83     assertEquals("", dbref.getVersion());
  84     assertEquals(2, ef.getQualifiers().size());
  85
  86     // feature qualifiers
  87     Qualifier q = ef.getQualifiers().get(0);
  88     assertEquals("note", q.getName());
  89     assertEquals(2, q.getValues().length);
  90     assertEquals("L-lactate dehydrogenase A-chain", q.getValues()[0]);
  91     assertEquals("pickle", q.getValues()[1]);
  92     assertNull(q.getEvidence());
  93     q = ef.getQualifiers().get(1);
  94     assertEquals("translation", q.getName());
  95     assertEquals(1, q.getValues().length);
  96     assertEquals("MSLKDHLIHN", q.getValues()[0]);
  97     assertEquals(1, q.getEvidence().length);
  98     assertEquals("Keith", q.getEvidence()[0]);
  99
 100     // feature locations
 101     assertEquals(1, ef.getLocations().size());
 102     EmblFeatureLocations fl = ef.getLocations().get(0);
 103     assertEquals("single", fl.getLocationType());
 104     assertTrue(fl.isLocationComplement());
 105     assertEquals(1, fl.getLocElements().size());
 106     EmblFeatureLocElement le = fl.getLocElements().get(0);
 107     assertEquals("range", le.getType());
 108     assertEquals("X53828", le.getAccession());
 109     assertEquals("1", le.getVersion());
 110     assertFalse(le.isComplement());
 111     assertEquals(2, le.getBasePositions().length);
 112     BasePosition bp = le.getBasePositions()[0];
 113     assertEquals("simple", bp.getType());
 114     assertEquals("60", bp.getPos());
 115     bp = le.getBasePositions()[1];
 116     assertEquals("join", bp.getType());
 117     assertEquals("1058", bp.getPos());
 118
 119     /*
 120      * Sequence
 121      */
 122     EmblSequence seq = entry.getSequence();
 123     assertEquals("mRNA", seq.getType());
 124     assertEquals("2", seq.getVersion());
 125     assertEquals("GTGACG", seq.getSequence());
 126   }
 127 }