test/jalview/ws/dbsources/UniprotTest.java

   1 /*
   2  * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
   3  * Copyright (C) $$Year-Rel$$ The Jalview Authors
   4  *
   5  * This file is part of Jalview.
   6  *
   7  * Jalview is free software: you can redistribute it and/or
   8  * modify it under the terms of the GNU General Public License
   9  * as published by the Free Software Foundation, either version 3
  10  * of the License, or (at your option) any later version.
  11  *
  12  * Jalview is distributed in the hope that it will be useful, but
  13  * WITHOUT ANY WARRANTY; without even the implied warranty
  14  * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
  15  * PURPOSE.  See the GNU General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU General Public License
  18  * along with Jalview.  If not, see <http://www.gnu.org/licenses/>.
  19  * The Jalview Authors are detailed in the 'AUTHORS' file.
  20  */
  21 package jalview.ws.dbsources;
  22
  23 import static org.testng.Assert.assertFalse;
  24 import static org.testng.AssertJUnit.assertEquals;
  25 import static org.testng.AssertJUnit.assertNotNull;
  26 import static org.testng.AssertJUnit.assertNull;
  27 import static org.testng.AssertJUnit.assertTrue;
  28
  29 import jalview.datamodel.DBRefEntry;
  30 import jalview.datamodel.DBRefSource;
  31 import jalview.datamodel.SequenceI;
  32 import jalview.gui.JvOptionPane;
  33 import jalview.util.DBRefUtils;
  34 import jalview.xml.binding.uniprot.DbReferenceType;
  35 import jalview.xml.binding.uniprot.Entry;
  36 import jalview.xml.binding.uniprot.FeatureType;
  37 import jalview.xml.binding.uniprot.LocationType;
  38 import jalview.xml.binding.uniprot.PositionType;
  39
  40 import java.io.ByteArrayInputStream;
  41 import java.io.InputStream;
  42 import java.io.UnsupportedEncodingException;
  43 import java.math.BigInteger;
  44 import java.util.List;
  45
  46 import org.testng.Assert;
  47 import org.testng.annotations.BeforeClass;
  48 import org.testng.annotations.Test;
  49
  50 public class UniprotTest
  51 {
  52
  53   @BeforeClass(alwaysRun = true)
  54   public void setUpJvOptionPane()
  55   {
  56     JvOptionPane.setInteractiveMode(false);
  57     JvOptionPane.setMockResponse(JvOptionPane.CANCEL_OPTION);
  58   }
  59
  60   // adapted from http://www.uniprot.org/uniprot/A9CKP4.xml
  61   private static final String UNIPROT_XML = "<?xml version='1.0' encoding='UTF-8'?>"
  62           + "<uniprot xmlns=\"http://uniprot.org/uniprot\">"
  63           + "<entry dataset=\"TrEMBL\" created=\"2008-01-15\" modified=\"2015-03-04\" version=\"38\">"
  64           + "<accession>A9CKP4</accession>"
  65           + "<accession>A9CKP5</accession>"
  66           + "<name>A9CKP4_AGRT5</name>"
  67           + "<name>A9CKP4_AGRT6</name>"
  68           + "<protein><recommendedName><fullName>Mitogen-activated protein kinase 13</fullName></recommendedName></protein>"
  69           + "<dbReference type=\"PDB\" id=\"2FSQ\"><property type=\"method\" value=\"X-ray\"/><property type=\"resolution\" value=\"1.40\"/></dbReference>"
  70           + "<dbReference type=\"PDBsum\" id=\"2FSR\"/>"
  71           + "<dbReference type=\"EMBL\" id=\"AE007869\"><property type=\"protein sequence ID\" value=\"AAK85932.1\"/><property type=\"molecule type\" value=\"Genomic_DNA\"/></dbReference>"
  72           + "<feature type=\"signal peptide\" evidence=\"7\"><location><begin position=\"1\"/><end position=\"18\"/></location></feature>"
  73           + "<feature type=\"propeptide\" description=\"Activation peptide\" id=\"PRO_0000027399\" evidence=\"9 16 17 18\"><location><begin position=\"19\"/><end position=\"20\"/></location></feature>"
  74           + "<feature type=\"chain\" description=\"Granzyme B\" id=\"PRO_0000027400\"><location><begin position=\"21\"/><end position=\"247\"/></location></feature>"
  75           + "<feature type=\"sequence variant\"><original>M</original><variation>L</variation><location><position position=\"41\"/></location></feature>"
  76           + "<feature type=\"sequence variant\" description=\"Pathogenic\"><original>M</original><variation>L</variation><location><position position=\"41\"/></location></feature>"
  77           + "<feature type=\"sequence variant\" description=\"Pathogenic\"><original>M</original><location><position position=\"41\"/></location></feature>"
  78           + "<feature type=\"sequence variant\" description=\"Foo\"><variation>L</variation><variation>LMV</variation><original>M</original><location><position position=\"42\"/></location></feature>"
  79           + "<feature type=\"sequence variant\" description=\"Foo\"><variation>LL</variation><variation>LMV</variation><original>ML</original><location><begin position=\"42\"/><end position=\"43\"/></location></feature>"
  80           + "<feature type=\"sequence variant\" description=\"Foo Too\"><variation>LL</variation><variation>LMVK</variation><original>MLML</original><location><begin position=\"42\"/><end position=\"45\"/></location></feature>"
  81           + "<sequence length=\"10\" mass=\"27410\" checksum=\"8CB760AACF88FE6C\" modified=\"2008-01-15\" version=\"1\">MHAPL VSKDL</sequence></entry>"
  82           + "</uniprot>";
  83
  84   /**
  85    * Test the method that unmarshals XML to a Uniprot model
  86    *
  87    * @throws UnsupportedEncodingException
  88    */
  89   @Test(groups = { "Functional" })
  90   public void testGetUniprotEntries() throws UnsupportedEncodingException
  91   {
  92     Uniprot u = new Uniprot();
  93     InputStream is = new ByteArrayInputStream(UNIPROT_XML.getBytes());
  94     List<Entry> entries = u.getUniprotEntries(is);
  95     assertEquals(1, entries.size());
  96     Entry entry = entries.get(0);
  97     assertEquals(2, entry.getName().size());
  98     assertEquals("A9CKP4_AGRT5", entry.getName().get(0));
  99     assertEquals("A9CKP4_AGRT6", entry.getName().get(1));
 100     assertEquals(2, entry.getAccession().size());
 101     assertEquals("A9CKP4", entry.getAccession().get(0));
 102     assertEquals("A9CKP5", entry.getAccession().get(1));
 103
 104     assertEquals("MHAPL VSKDL", entry.getSequence().getValue());
 105
 106     assertEquals("Mitogen-activated protein kinase 13", entry.getProtein()
 107             .getRecommendedName().getFullName().getValue());
 108
 109     /*
 110      * Check sequence features
 111      */
 112     List<FeatureType> features = entry.getFeature();
 113     assertEquals(9, features.size());
 114     FeatureType sf = features.get(0);
 115     assertEquals("signal peptide", sf.getType());
 116     assertNull(sf.getDescription());
 117     assertNull(sf.getStatus());
 118     assertNull(sf.getLocation().getPosition());
 119     assertEquals(1, sf.getLocation().getBegin().getPosition().intValue());
 120     assertEquals(18, sf.getLocation().getEnd().getPosition().intValue());
 121     sf = features.get(1);
 122     assertEquals("propeptide", sf.getType());
 123     assertEquals("Activation peptide", sf.getDescription());
 124     assertNull(sf.getLocation().getPosition());
 125     assertEquals(19, sf.getLocation().getBegin().getPosition().intValue());
 126     assertEquals(20, sf.getLocation().getEnd().getPosition().intValue());
 127     sf = features.get(2);
 128     assertEquals("chain", sf.getType());
 129     assertEquals("Granzyme B", sf.getDescription());
 130     assertNull(sf.getLocation().getPosition());
 131     assertEquals(21, sf.getLocation().getBegin().getPosition().intValue());
 132     assertEquals(247, sf.getLocation().getEnd().getPosition().intValue());
 133
 134     sf = features.get(3);
 135     assertEquals("sequence variant", sf.getType());
 136     assertNull(sf.getDescription());
 137     assertEquals(41,
 138             sf.getLocation().getPosition().getPosition().intValue());
 139     assertNull(sf.getLocation().getBegin());
 140     assertNull(sf.getLocation().getEnd());
 141
 142     sf = features.get(4);
 143     assertEquals("sequence variant", sf.getType());
 144     assertEquals("Pathogenic", sf.getDescription());
 145     assertEquals(41,
 146             sf.getLocation().getPosition().getPosition().intValue());
 147     assertNull(sf.getLocation().getBegin());
 148     assertNull(sf.getLocation().getEnd());
 149
 150     sf = features.get(5);
 151     assertEquals("sequence variant", sf.getType());
 152     assertEquals("Pathogenic", sf.getDescription());
 153     assertEquals(41,
 154             sf.getLocation().getPosition().getPosition().intValue());
 155     assertNull(sf.getLocation().getBegin());
 156     assertNull(sf.getLocation().getEnd());
 157
 158     sf = features.get(6);
 159     assertEquals("sequence variant", sf.getType());
 160     assertEquals("Foo",
 161             sf.getDescription());
 162     assertEquals(42,
 163             sf.getLocation().getPosition().getPosition().intValue());
 164     assertNull(sf.getLocation().getBegin());
 165     assertNull(sf.getLocation().getEnd());
 166     Assert.assertEquals(Uniprot.getDescription(sf),
 167             "<html>p.Met42Leu" + "<br/>&nbsp;&nbsp;"
 168                     + "p.Met42LeuMetVal Foo</html>");
 169
 170     sf = features.get(7);
 171     assertNull(sf.getLocation().getPosition());
 172     assertEquals(42, sf.getLocation().getBegin().getPosition().intValue());
 173     assertEquals(43, sf.getLocation().getEnd().getPosition().intValue());
 174     Assert.assertEquals(Uniprot.getDescription(sf),
 175             "<html>p.MetLeu42LeuLeu" + "<br/>&nbsp;&nbsp;"
 176                     + "p.MetLeu42LeuMetVal Foo</html>");
 177
 178     sf = features.get(8);
 179     assertNull(sf.getLocation().getPosition());
 180     assertEquals(42, sf.getLocation().getBegin().getPosition().intValue());
 181     assertEquals(45, sf.getLocation().getEnd().getPosition().intValue());
 182     Assert.assertEquals(Uniprot.getDescription(sf),
 183             "<html>p.MLML42LeuLeu" + "<br/>&nbsp;&nbsp;"
 184                     + "p.MLML42LMVK Foo Too</html>");
 185
 186     /*
 187      * Check cross-references
 188      */
 189     List<DbReferenceType> xrefs = entry.getDbReference();
 190     assertEquals(3, xrefs.size());
 191
 192     DbReferenceType xref = xrefs.get(0);
 193     assertEquals("2FSQ", xref.getId());
 194     assertEquals("PDB", xref.getType());
 195     assertEquals("X-ray",
 196             Uniprot.getProperty(xref.getProperty(), "method"));
 197     assertEquals("1.40",
 198             Uniprot.getProperty(xref.getProperty(), "resolution"));
 199
 200     xref = xrefs.get(1);
 201     assertEquals("2FSR", xref.getId());
 202     assertEquals("PDBsum", xref.getType());
 203     assertTrue(xref.getProperty().isEmpty());
 204
 205     xref = xrefs.get(2);
 206     assertEquals("AE007869", xref.getId());
 207     assertEquals("EMBL", xref.getType());
 208     assertEquals("AAK85932.1",
 209             Uniprot.getProperty(xref.getProperty(), "protein sequence ID"));
 210     assertEquals("Genomic_DNA",
 211             Uniprot.getProperty(xref.getProperty(), "molecule type"));
 212   }
 213
 214   @Test(groups = { "Functional" })
 215   public void testGetUniprotSequence() throws UnsupportedEncodingException
 216   {
 217     InputStream is = new ByteArrayInputStream(UNIPROT_XML.getBytes());
 218     Entry entry = new Uniprot().getUniprotEntries(
 219             is).get(0);
 220     SequenceI seq = new Uniprot().uniprotEntryToSequence(entry);
 221     assertNotNull(seq);
 222     assertEquals(6, seq.getDBRefs().size()); // 2*Uniprot, PDB, PDBsum, 2*EMBL
 223     assertEquals(seq.getSequenceAsString(),
 224             seq.createDatasetSequence().getSequenceAsString());
 225     assertEquals(2,seq.getPrimaryDBRefs().size());
 226     List<DBRefEntry> res = DBRefUtils.searchRefs(seq.getPrimaryDBRefs(), "A9CKP4");
 227     assertEquals(1,res.size());
 228     assertTrue(res.get(0).isCanonical());
 229     res = DBRefUtils.searchRefsForSource(seq.getDBRefs(), DBRefSource.UNIPROT);
 230     assertEquals(2,res.size());
 231     /*
 232      * NB this test fragile - relies on ordering being preserved
 233      */
 234     assertTrue(res.get(0).isCanonical());
 235     assertFalse(res.get(1).isCanonical());
 236   }
 237
 238   /**
 239    * Test the method that formats the sequence id
 240    *
 241    * @throws UnsupportedEncodingException
 242    */
 243   @Test(groups = { "Functional" })
 244   public void testGetUniprotEntryId() throws UnsupportedEncodingException
 245   {
 246     InputStream is = new ByteArrayInputStream(UNIPROT_XML.getBytes());
 247     Entry entry = new Uniprot().getUniprotEntries(is).get(0);
 248
 249     /*
 250      * name formatted with Uniprot Entry name
 251      */
 252     String expectedName = "A9CKP4_AGRT5|A9CKP4_AGRT6";
 253     assertEquals(expectedName,
 254             Uniprot.getUniprotEntryId(entry));
 255   }
 256
 257   /**
 258    * Test the method that formats the sequence description
 259    *
 260    * @throws UnsupportedEncodingException
 261    */
 262   @Test(groups = { "Functional" })
 263   public void testGetUniprotEntryDescription()
 264           throws UnsupportedEncodingException
 265   {
 266     InputStream is = new ByteArrayInputStream(UNIPROT_XML.getBytes());
 267     Entry entry = new Uniprot().getUniprotEntries(is).get(0);
 268
 269     assertEquals("Mitogen-activated protein kinase 13",
 270             Uniprot.getUniprotEntryDescription(entry));
 271   }
 272
 273   @Test(groups = { "Functional" })
 274   public void testGetDescription()
 275   {
 276     FeatureType ft = new FeatureType();
 277     assertEquals("", Uniprot.getDescription(ft));
 278
 279     ft.setDescription("Hello");
 280     assertEquals("Hello", Uniprot.getDescription(ft));
 281
 282     ft.setLocation(new LocationType());
 283     ft.getLocation().setPosition(new PositionType());
 284     ft.getLocation().getPosition().setPosition(BigInteger.valueOf(23));
 285     ft.setOriginal("K");
 286     ft.getVariation().add("y");
 287     assertEquals("p.Lys23Tyr Hello", Uniprot.getDescription(ft));
 288
 289     // multiple variants generate an html description over more than one line
 290     ft.getVariation().add("W");
 291     assertEquals("<html>p.Lys23Tyr<br/>&nbsp;&nbsp;p.Lys23Trp Hello</html>",
 292             Uniprot.getDescription(ft));
 293
 294     /*
 295      * indel cases
 296      * up to 3 bases (original or variant) are shown using 3 letter code
 297      */
 298     ft.getVariation().clear();
 299     ft.getVariation().add("KWE");
 300     ft.setOriginal("KLS");
 301     assertEquals("p.LysLeuSer23LysTrpGlu Hello",
 302             Uniprot.getDescription(ft));
 303
 304     // adding a fourth original base switches to single letter code
 305     ft.setOriginal("KLST");
 306     assertEquals("p.KLST23LysTrpGlu Hello", Uniprot.getDescription(ft));
 307
 308     // adding a fourth variant switches to single letter code
 309     ft.getVariation().clear();
 310     ft.getVariation().add("KWES");
 311     assertEquals("p.KLST23KWES Hello", Uniprot.getDescription(ft));
 312
 313     ft.getVariation().clear();
 314     ft.getVariation().add("z"); // unknown variant - fails gracefully
 315     ft.setOriginal("K");
 316     assertEquals("p.Lys23z Hello", Uniprot.getDescription(ft));
 317
 318     ft.getVariation().clear(); // variant missing - is ignored
 319     assertEquals("Hello", Uniprot.getDescription(ft));
 320   }
 321 }