test/jalview/io/FeaturesFileTest.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.io;
  22
  23 import static org.testng.AssertJUnit.assertEquals;
  24 import static org.testng.AssertJUnit.assertFalse;
  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.api.FeatureColourI;
  30 import jalview.api.FeatureRenderer;
  31 import jalview.datamodel.Alignment;
  32 import jalview.datamodel.AlignmentI;
  33 import jalview.datamodel.SequenceDummy;
  34 import jalview.datamodel.SequenceFeature;
  35 import jalview.datamodel.SequenceI;
  36 import jalview.gui.AlignFrame;
  37
  38 import java.awt.Color;
  39 import java.io.File;
  40 import java.io.IOException;
  41 import java.util.Map;
  42
  43 import org.testng.annotations.Test;
  44
  45 public class FeaturesFileTest
  46 {
  47
  48   private static String simpleGffFile = "examples/testdata/simpleGff3.gff";
  49
  50   @Test(groups = { "Functional" })
  51   public void testParse() throws Exception
  52   {
  53     File f = new File("examples/uniref50.fa");
  54     AlignmentI al = readAlignmentFile(f);
  55     AlignFrame af = new AlignFrame(al, 500, 500);
  56     Map<String, FeatureColourI> colours = af.getFeatureRenderer()
  57             .getFeatureColours();
  58     FeaturesFile featuresFile = new FeaturesFile(
  59             "examples/exampleFeatures.txt", FormatAdapter.FILE);
  60     assertTrue("Test " + "Features file test"
  61             + "\nFailed to parse features file.",
  62             featuresFile.parse(al.getDataset(), colours, true));
  63
  64     /*
  65      * Refetch the colour map from the FeatureRenderer (to confirm it has been
  66      * updated - JAL-1904), and verify (some) feature group colours
  67      */
  68     colours = af.getFeatureRenderer().getFeatureColours();
  69     assertEquals("26 feature group colours not found", 26, colours.size());
  70     assertEquals(colours.get("Cath").getColour(), new Color(0x93b1d1));
  71     assertEquals(colours.get("ASX-MOTIF").getColour(), new Color(0x6addbb));
  72
  73     /*
  74      * verify (some) features on sequences
  75      */
  76     SequenceFeature[] sfs = al.getSequenceAt(0).getDatasetSequence()
  77             .getSequenceFeatures(); // FER_CAPAA
  78     assertEquals(7, sfs.length);
  79     SequenceFeature sf = sfs[0];
  80     assertEquals("Iron-sulfur (2Fe-2S)", sf.description);
  81     assertEquals(39, sf.begin);
  82     assertEquals(39, sf.end);
  83     assertEquals("uniprot", sf.featureGroup);
  84     assertEquals("METAL", sf.type);
  85     sf = sfs[1];
  86     assertEquals("Iron-sulfur (2Fe-2S)", sf.description);
  87     assertEquals(44, sf.begin);
  88     assertEquals(44, sf.end);
  89     assertEquals("uniprot", sf.featureGroup);
  90     assertEquals("METAL", sf.type);
  91     sf = sfs[2];
  92     assertEquals("Iron-sulfur (2Fe-2S)", sf.description);
  93     assertEquals(47, sf.begin);
  94     assertEquals(47, sf.end);
  95     assertEquals("uniprot", sf.featureGroup);
  96     assertEquals("METAL", sf.type);
  97     sf = sfs[3];
  98     assertEquals("Iron-sulfur (2Fe-2S)", sf.description);
  99     assertEquals(77, sf.begin);
 100     assertEquals(77, sf.end);
 101     assertEquals("uniprot", sf.featureGroup);
 102     assertEquals("METAL", sf.type);
 103     sf = sfs[4];
 104     assertEquals("Fer2 Status: True Positive Pfam 8_8%LINK%",
 105             sf.description);
 106     assertEquals("Pfam 8_8|http://pfam.sanger.ac.uk/family/PF00111",
 107             sf.links.get(0).toString());
 108     assertEquals(8, sf.begin);
 109     assertEquals(83, sf.end);
 110     assertEquals("uniprot", sf.featureGroup);
 111     assertEquals("Pfam", sf.type);
 112     sf = sfs[5];
 113     assertEquals("Ferredoxin_fold Status: True Positive ", sf.description);
 114     assertEquals(3, sf.begin);
 115     assertEquals(93, sf.end);
 116     assertEquals("uniprot", sf.featureGroup);
 117     assertEquals("Cath", sf.type);
 118     sf = sfs[6];
 119     assertEquals(
 120             "High confidence server. Only hits with scores over 0.8 are reported. PHOSPHORYLATION (T) 89_8%LINK%",
 121             sf.description);
 122     assertEquals(
 123             "PHOSPHORYLATION (T) 89_8|http://www.cbs.dtu.dk/cgi-bin/proview/webface-link?seqid=P83527&amp;service=NetPhos-2.0",
 124             sf.links.get(0).toString());
 125     assertEquals(89, sf.begin);
 126     assertEquals(89, sf.end);
 127     assertEquals("netphos", sf.featureGroup);
 128     assertEquals("PHOSPHORYLATION (T)", sf.type);
 129   }
 130
 131   /**
 132    * Test parsing a features file with a mix of Jalview and GFF formatted
 133    * content
 134    *
 135    * @throws Exception
 136    */
 137   @Test(groups = { "Functional" })
 138   public void testParse_mixedJalviewGff() throws Exception
 139   {
 140     File f = new File("examples/uniref50.fa");
 141     AlignmentI al = readAlignmentFile(f);
 142     AlignFrame af = new AlignFrame(al, 500, 500);
 143     Map<String, FeatureColourI> colours = af.getFeatureRenderer()
 144             .getFeatureColours();
 145     // GFF2 uses space as name/value separator in column 9
 146     String gffData = "METAL\tcc9900\n" + "GFF\n"
 147             + "FER_CAPAA\tuniprot\tMETAL\t44\t45\t4.0\t.\t.\tNote Iron-sulfur; Note 2Fe-2S\n"
 148             + "FER1_SOLLC\tuniprot\tPfam\t55\t130\t2.0\t.\t.";
 149     FeaturesFile featuresFile = new FeaturesFile(gffData,
 150             FormatAdapter.PASTE);
 151     assertTrue("Failed to parse features file",
 152             featuresFile.parse(al.getDataset(), colours, true));
 153
 154     // verify colours read or synthesized
 155     colours = af.getFeatureRenderer().getFeatureColours();
 156     assertEquals("1 feature group colours not found", 1, colours.size());
 157     assertEquals(colours.get("METAL").getColour(), new Color(0xcc9900));
 158
 159     // verify feature on FER_CAPAA
 160     SequenceFeature[] sfs = al.getSequenceAt(0).getDatasetSequence()
 161             .getSequenceFeatures();
 162     assertEquals(1, sfs.length);
 163     SequenceFeature sf = sfs[0];
 164     assertEquals("Iron-sulfur,2Fe-2S", sf.description);
 165     assertEquals(44, sf.begin);
 166     assertEquals(45, sf.end);
 167     assertEquals("uniprot", sf.featureGroup);
 168     assertEquals("METAL", sf.type);
 169     assertEquals(4f, sf.getScore(), 0.001f);
 170
 171     // verify feature on FER1_SOLLC
 172     sfs = al.getSequenceAt(2).getDatasetSequence().getSequenceFeatures();
 173     assertEquals(1, sfs.length);
 174     sf = sfs[0];
 175     assertEquals("uniprot", sf.description);
 176     assertEquals(55, sf.begin);
 177     assertEquals(130, sf.end);
 178     assertEquals("uniprot", sf.featureGroup);
 179     assertEquals("Pfam", sf.type);
 180     assertEquals(2f, sf.getScore(), 0.001f);
 181   }
 182
 183   public static AlignmentI readAlignmentFile(File f) throws IOException
 184   {
 185     System.out.println("Reading file: " + f);
 186     String ff = f.getPath();
 187     FormatAdapter rf = new FormatAdapter();
 188
 189     AlignmentI al = rf.readFile(ff, FormatAdapter.FILE,
 190             new IdentifyFile().identify(ff, FormatAdapter.FILE));
 191
 192     al.setDataset(null); // creates dataset sequences
 193     assertNotNull("Couldn't read supplied alignment data.", al);
 194     return al;
 195   }
 196
 197   /**
 198    * Test parsing a features file with GFF formatted content only
 199    *
 200    * @throws Exception
 201    */
 202   @Test(groups = { "Functional" })
 203   public void testParse_pureGff3() throws Exception
 204   {
 205     File f = new File("examples/uniref50.fa");
 206     AlignmentI al = readAlignmentFile(f);
 207     AlignFrame af = new AlignFrame(al, 500, 500);
 208     Map<String, FeatureColourI> colours = af.getFeatureRenderer()
 209             .getFeatureColours();
 210     // GFF3 uses '=' separator for name/value pairs in colum 9
 211     String gffData = "##gff-version 3\n"
 212             + "FER_CAPAA\tuniprot\tMETAL\t39\t39\t0.0\t.\t.\t"
 213             + "Note=Iron-sulfur (2Fe-2S);Note=another note;evidence=ECO:0000255|PROSITE-ProRule:PRU00465\n"
 214             + "FER1_SOLLC\tuniprot\tPfam\t55\t130\t3.0\t.\t.\tID=$23";
 215     FeaturesFile featuresFile = new FeaturesFile(gffData,
 216             FormatAdapter.PASTE);
 217     assertTrue("Failed to parse features file",
 218             featuresFile.parse(al.getDataset(), colours, true));
 219
 220     // verify feature on FER_CAPAA
 221     SequenceFeature[] sfs = al.getSequenceAt(0).getDatasetSequence()
 222             .getSequenceFeatures();
 223     assertEquals(1, sfs.length);
 224     SequenceFeature sf = sfs[0];
 225     // description parsed from Note attribute
 226     assertEquals("Iron-sulfur (2Fe-2S),another note", sf.description);
 227     assertEquals(39, sf.begin);
 228     assertEquals(39, sf.end);
 229     assertEquals("uniprot", sf.featureGroup);
 230     assertEquals("METAL", sf.type);
 231     assertEquals(
 232             "Note=Iron-sulfur (2Fe-2S);Note=another note;evidence=ECO:0000255|PROSITE-ProRule:PRU00465",
 233             sf.getValue("ATTRIBUTES"));
 234
 235     // verify feature on FER1_SOLLC1
 236     sfs = al.getSequenceAt(2).getDatasetSequence().getSequenceFeatures();
 237     assertEquals(1, sfs.length);
 238     sf = sfs[0];
 239     // ID used for description if available
 240     assertEquals("$23", sf.description);
 241     assertEquals(55, sf.begin);
 242     assertEquals(130, sf.end);
 243     assertEquals("uniprot", sf.featureGroup);
 244     assertEquals("Pfam", sf.type);
 245     assertEquals(3f, sf.getScore(), 0.001f);
 246   }
 247
 248   /**
 249    * Test parsing a features file with Jalview format features (but no colour
 250    * descriptors or startgroup to give the hint not to parse as GFF)
 251    *
 252    * @throws Exception
 253    */
 254   @Test(groups = { "Functional" })
 255   public void testParse_jalviewFeaturesOnly() throws Exception
 256   {
 257     File f = new File("examples/uniref50.fa");
 258     AlignmentI al = readAlignmentFile(f);
 259     AlignFrame af = new AlignFrame(al, 500, 500);
 260     Map<String, FeatureColourI> colours = af.getFeatureRenderer()
 261             .getFeatureColours();
 262
 263     /*
 264      * one feature on FER_CAPAA and one on sequence 3 (index 2) FER1_SOLLC
 265      */
 266     String featureData = "Iron-sulfur (2Fe-2S)\tFER_CAPAA\t-1\t39\t39\tMETAL\n"
 267             + "Iron-phosphorus (2Fe-P)\tID_NOT_SPECIFIED\t2\t86\t87\tMETALLIC\n";
 268     FeaturesFile featuresFile = new FeaturesFile(featureData,
 269             FormatAdapter.PASTE);
 270     assertTrue("Failed to parse features file",
 271             featuresFile.parse(al.getDataset(), colours, true));
 272
 273     // verify FER_CAPAA feature
 274     SequenceFeature[] sfs = al.getSequenceAt(0).getDatasetSequence()
 275             .getSequenceFeatures();
 276     assertEquals(1, sfs.length);
 277     SequenceFeature sf = sfs[0];
 278     assertEquals("Iron-sulfur (2Fe-2S)", sf.description);
 279     assertEquals(39, sf.begin);
 280     assertEquals(39, sf.end);
 281     assertEquals("METAL", sf.type);
 282
 283     // verify FER1_SOLLC feature
 284     sfs = al.getSequenceAt(2).getDatasetSequence().getSequenceFeatures();
 285     assertEquals(1, sfs.length);
 286     sf = sfs[0];
 287     assertEquals("Iron-phosphorus (2Fe-P)", sf.description);
 288     assertEquals(86, sf.begin);
 289     assertEquals(87, sf.end);
 290     assertEquals("METALLIC", sf.type);
 291   }
 292
 293   private void checkDatasetfromSimpleGff3(AlignmentI dataset)
 294   {
 295     assertEquals("no sequences extracted from GFF3 file", 2,
 296             dataset.getHeight());
 297
 298     SequenceI seq1 = dataset.findName("seq1");
 299     SequenceI seq2 = dataset.findName("seq2");
 300     assertNotNull(seq1);
 301     assertNotNull(seq2);
 302     assertFalse(
 303             "Failed to replace dummy seq1 with real sequence",
 304             seq1 instanceof SequenceDummy
 305                     && ((SequenceDummy) seq1).isDummy());
 306     assertFalse(
 307             "Failed to replace dummy seq2 with real sequence",
 308             seq2 instanceof SequenceDummy
 309                     && ((SequenceDummy) seq2).isDummy());
 310     String placeholderseq = new SequenceDummy("foo").getSequenceAsString();
 311     assertFalse("dummy replacement buggy for seq1",
 312             placeholderseq.equals(seq1.getSequenceAsString()));
 313     assertFalse("dummy replacement buggy for seq2",
 314             placeholderseq.equals(seq2.getSequenceAsString()));
 315     assertNotNull("No features added to seq1", seq1.getSequenceFeatures());
 316     assertEquals("Wrong number of features", 3,
 317             seq1.getSequenceFeatures().length);
 318     assertNull(seq2.getSequenceFeatures());
 319     assertEquals(
 320             "Wrong number of features",
 321             0,
 322             seq2.getSequenceFeatures() == null ? 0 : seq2
 323                     .getSequenceFeatures().length);
 324     assertTrue(
 325             "Expected at least one CDNA/Protein mapping for seq1",
 326             dataset.getCodonFrame(seq1) != null
 327                     && dataset.getCodonFrame(seq1).size() > 0);
 328
 329   }
 330
 331   @Test(groups = { "Functional" })
 332   public void readGff3File() throws IOException
 333   {
 334     FeaturesFile gffreader = new FeaturesFile(true, simpleGffFile,
 335             FormatAdapter.FILE);
 336     Alignment dataset = new Alignment(gffreader.getSeqsAsArray());
 337     gffreader.addProperties(dataset);
 338     checkDatasetfromSimpleGff3(dataset);
 339   }
 340
 341   @Test(groups = { "Functional" })
 342   public void simpleGff3FileClass() throws IOException
 343   {
 344     AlignmentI dataset = new Alignment(new SequenceI[] {});
 345     FeaturesFile ffile = new FeaturesFile(simpleGffFile,
 346             FormatAdapter.FILE);
 347
 348     boolean parseResult = ffile.parse(dataset, null, false, false);
 349     assertTrue("return result should be true", parseResult);
 350     checkDatasetfromSimpleGff3(dataset);
 351   }
 352
 353   @Test(groups = { "Functional" })
 354   public void simpleGff3FileLoader() throws IOException
 355   {
 356     AlignFrame af = new FileLoader(false).LoadFileWaitTillLoaded(
 357             simpleGffFile, FormatAdapter.FILE);
 358     assertTrue(
 359             "Didn't read the alignment into an alignframe from Gff3 File",
 360             af != null);
 361     checkDatasetfromSimpleGff3(af.getViewport().getAlignment());
 362   }
 363
 364   @Test(groups = { "Functional" })
 365   public void simpleGff3RelaxedIdMatching() throws IOException
 366   {
 367     AlignmentI dataset = new Alignment(new SequenceI[] {});
 368     FeaturesFile ffile = new FeaturesFile(simpleGffFile,
 369             FormatAdapter.FILE);
 370
 371     boolean parseResult = ffile.parse(dataset, null, false, true);
 372     assertTrue("return result (relaxedID matching) should be true",
 373             parseResult);
 374     checkDatasetfromSimpleGff3(dataset);
 375   }
 376
 377   @Test(groups = { "Functional" })
 378   public void testPrintJalviewFormat() throws Exception
 379   {
 380     File f = new File("examples/uniref50.fa");
 381     AlignmentI al = readAlignmentFile(f);
 382     AlignFrame af = new AlignFrame(al, 500, 500);
 383     Map<String, FeatureColourI> colours = af.getFeatureRenderer()
 384             .getFeatureColours();
 385     String features = "METAL\tcc9900\n"
 386             + "GAMMA-TURN\tred|0,255,255|20.0|95.0|below|66.0\n"
 387             + "Pfam\tred\n"
 388             + "STARTGROUP\tuniprot\n"
 389             + "Iron\tFER_CAPAA\t-1\t39\t39\tMETAL\n"
 390             + "Turn\tFER_CAPAA\t-1\t36\t38\tGAMMA-TURN\n"
 391             + "<html>Pfam domain<a href=\"http://pfam.sanger.ac.uk/family/PF00111\">Pfam_3_4</a></html>\tFER_CAPAA\t-1\t20\t20\tPfam\n"
 392             + "ENDGROUP\tuniprot\n";
 393     FeaturesFile featuresFile = new FeaturesFile(features,
 394             FormatAdapter.PASTE);
 395     featuresFile.parse(al.getDataset(), colours, false);
 396
 397     /*
 398      * first with no features displayed
 399      */
 400     FeatureRenderer fr = af.alignPanel.getFeatureRenderer();
 401     Map<String, FeatureColourI> visible = fr
 402             .getDisplayedFeatureCols();
 403     String exported = featuresFile.printJalviewFormat(
 404             al.getSequencesArray(), visible);
 405     String expected = "No Features Visible";
 406     assertEquals(expected, exported);
 407
 408     /*
 409      * set METAL (in uniprot group) and GAMMA-TURN visible, but not Pfam
 410      */
 411     fr.setVisible("METAL");
 412     fr.setVisible("GAMMA-TURN");
 413     visible = fr.getDisplayedFeatureCols();
 414     exported = featuresFile.printJalviewFormat(al.getSequencesArray(),
 415             visible);
 416     expected = "METAL\tcc9900\n"
 417             + "GAMMA-TURN\tff0000|00ffff|20.0|95.0|below|66.0\n"
 418             + "\nSTARTGROUP\tuniprot\n"
 419             + "Iron\tFER_CAPAA\t-1\t39\t39\tMETAL\t0.0\n"
 420             + "Turn\tFER_CAPAA\t-1\t36\t38\tGAMMA-TURN\t0.0\n"
 421             + "ENDGROUP\tuniprot\n";
 422     assertEquals(expected, exported);
 423
 424     /*
 425      * now set Pfam visible
 426      */
 427     fr.setVisible("Pfam");
 428     visible = fr.getDisplayedFeatureCols();
 429     exported = featuresFile.printJalviewFormat(al.getSequencesArray(),
 430             visible);
 431     /*
 432      * note the order of feature types is uncontrolled - derives from
 433      * FeaturesDisplayed.featuresDisplayed which is a HashSet
 434      */
 435     expected = "METAL\tcc9900\n"
 436             + "Pfam\tff0000\n"
 437             + "GAMMA-TURN\tff0000|00ffff|20.0|95.0|below|66.0\n"
 438             + "\nSTARTGROUP\tuniprot\n"
 439             + "Iron\tFER_CAPAA\t-1\t39\t39\tMETAL\t0.0\n"
 440             + "Turn\tFER_CAPAA\t-1\t36\t38\tGAMMA-TURN\t0.0\n"
 441             + "<html>Pfam domain<a href=\"http://pfam.sanger.ac.uk/family/PF00111\">Pfam_3_4</a></html>\tFER_CAPAA\t-1\t20\t20\tPfam\t0.0\n"
 442             + "ENDGROUP\tuniprot\n";
 443     assertEquals(expected, exported);
 444   }
 445 }