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.FeatureRenderer;
  30 import jalview.datamodel.Alignment;
  31 import jalview.datamodel.AlignmentI;
  32 import jalview.datamodel.SequenceDummy;
  33 import jalview.datamodel.SequenceFeature;
  34 import jalview.datamodel.SequenceI;
  35 import jalview.gui.AlignFrame;
  36 import jalview.schemes.AnnotationColourGradient;
  37 import jalview.schemes.GraduatedColor;
  38
  39 import java.awt.Color;
  40 import java.io.File;
  41 import java.io.IOException;
  42 import java.util.Map;
  43
  44 import org.testng.annotations.Test;
  45
  46 public class FeaturesFileTest
  47 {
  48
  49   private static String simpleGffFile = "examples/testdata/simpleGff3.gff";
  50
  51   @Test(groups = { "Functional" })
  52   public void testParse() throws Exception
  53   {
  54     File f = new File("examples/uniref50.fa");
  55     AlignmentI al = readAlignmentFile(f);
  56     AlignFrame af = new AlignFrame(al, 500, 500);
  57     Map<String, Object> colours = af.getFeatureRenderer()
  58             .getFeatureColours();
  59     FeaturesFile featuresFile = new FeaturesFile(
  60             "examples/exampleFeatures.txt", FormatAdapter.FILE);
  61     assertTrue("Test " + "Features file test"
  62             + "\nFailed to parse features file.",
  63             featuresFile.parse(al.getDataset(), colours, true));
  64
  65     /*
  66      * Refetch the colour map from the FeatureRenderer (to confirm it has been
  67      * updated - JAL-1904), and verify (some) feature group colours
  68      */
  69     colours = af.getFeatureRenderer().getFeatureColours();
  70     assertEquals("26 feature group colours not found", 26, colours.size());
  71     assertEquals(colours.get("Cath"), new Color(0x93b1d1));
  72     assertEquals(colours.get("ASX-MOTIF"), new Color(0x6addbb));
  73
  74     /*
  75      * verify (some) features on sequences
  76      */
  77     SequenceFeature[] sfs = al.getSequenceAt(0).getDatasetSequence()
  78             .getSequenceFeatures(); // FER_CAPAA
  79     assertEquals(7, sfs.length);
  80     SequenceFeature sf = sfs[0];
  81     assertEquals("Iron-sulfur (2Fe-2S)", sf.description);
  82     assertEquals(39, sf.begin);
  83     assertEquals(39, sf.end);
  84     assertEquals("uniprot", sf.featureGroup);
  85     assertEquals("METAL", sf.type);
  86     sf = sfs[1];
  87     assertEquals("Iron-sulfur (2Fe-2S)", sf.description);
  88     assertEquals(44, sf.begin);
  89     assertEquals(44, sf.end);
  90     assertEquals("uniprot", sf.featureGroup);
  91     assertEquals("METAL", sf.type);
  92     sf = sfs[2];
  93     assertEquals("Iron-sulfur (2Fe-2S)", sf.description);
  94     assertEquals(47, sf.begin);
  95     assertEquals(47, sf.end);
  96     assertEquals("uniprot", sf.featureGroup);
  97     assertEquals("METAL", sf.type);
  98     sf = sfs[3];
  99     assertEquals("Iron-sulfur (2Fe-2S)", sf.description);
 100     assertEquals(77, sf.begin);
 101     assertEquals(77, sf.end);
 102     assertEquals("uniprot", sf.featureGroup);
 103     assertEquals("METAL", sf.type);
 104     sf = sfs[4];
 105     assertEquals("Fer2 Status: True Positive Pfam 8_8%LINK%",
 106             sf.description);
 107     assertEquals("Pfam 8_8|http://pfam.xfam.org/family/PF00111",
 108             sf.links.get(0).toString());
 109     assertEquals(8, sf.begin);
 110     assertEquals(83, sf.end);
 111     assertEquals("uniprot", sf.featureGroup);
 112     assertEquals("Pfam", sf.type);
 113     sf = sfs[5];
 114     assertEquals("Ferredoxin_fold Status: True Positive ", sf.description);
 115     assertEquals(3, sf.begin);
 116     assertEquals(93, sf.end);
 117     assertEquals("uniprot", sf.featureGroup);
 118     assertEquals("Cath", sf.type);
 119     sf = sfs[6];
 120     assertEquals(
 121             "High confidence server. Only hits with scores over 0.8 are reported. PHOSPHORYLATION (T) 89_8%LINK%",
 122             sf.description);
 123     assertEquals(
 124             "PHOSPHORYLATION (T) 89_8|http://www.cbs.dtu.dk/cgi-bin/proview/webface-link?seqid=P83527&amp;service=NetPhos-2.0",
 125             sf.links.get(0).toString());
 126     assertEquals(89, sf.begin);
 127     assertEquals(89, sf.end);
 128     assertEquals("netphos", sf.featureGroup);
 129     assertEquals("PHOSPHORYLATION (T)", sf.type);
 130   }
 131
 132   /**
 133    * Test parsing a features file with a mix of Jalview and GFF formatted
 134    * content
 135    *
 136    * @throws Exception
 137    */
 138   @Test(groups = { "Functional" })
 139   public void testParse_mixedJalviewGff() throws Exception
 140   {
 141     File f = new File("examples/uniref50.fa");
 142     AlignmentI al = readAlignmentFile(f);
 143     AlignFrame af = new AlignFrame(al, 500, 500);
 144     Map<String, Object> colours = af.getFeatureRenderer()
 145             .getFeatureColours();
 146     // GFF2 uses space as name/value separator in column 9
 147     String gffData = "METAL\tcc9900\n" + "GFF\n"
 148             + "FER_CAPAA\tuniprot\tMETAL\t44\t45\t4.0\t.\t.\tNote Iron-sulfur; Note 2Fe-2S\n"
 149             + "FER1_SOLLC\tuniprot\tPfam\t55\t130\t2.0\t.\t.";
 150     FeaturesFile featuresFile = new FeaturesFile(gffData,
 151             FormatAdapter.PASTE);
 152     assertTrue("Failed to parse features file",
 153             featuresFile.parse(al.getDataset(), colours, true));
 154
 155     // verify colours read or synthesized
 156     colours = af.getFeatureRenderer().getFeatureColours();
 157     assertEquals("1 feature group colours not found", 1, colours.size());
 158     assertEquals(colours.get("METAL"), new Color(0xcc9900));
 159
 160     // verify feature on FER_CAPAA
 161     SequenceFeature[] sfs = al.getSequenceAt(0).getDatasetSequence()
 162             .getSequenceFeatures();
 163     assertEquals(1, sfs.length);
 164     SequenceFeature sf = sfs[0];
 165     assertEquals("Iron-sulfur,2Fe-2S", sf.description);
 166     assertEquals(44, sf.begin);
 167     assertEquals(45, sf.end);
 168     assertEquals("uniprot", sf.featureGroup);
 169     assertEquals("METAL", sf.type);
 170     assertEquals(4f, sf.getScore(), 0.001f);
 171
 172     // verify feature on FER1_SOLLC
 173     sfs = al.getSequenceAt(2).getDatasetSequence().getSequenceFeatures();
 174     assertEquals(1, sfs.length);
 175     sf = sfs[0];
 176     assertEquals("uniprot", sf.description);
 177     assertEquals(55, sf.begin);
 178     assertEquals(130, sf.end);
 179     assertEquals("uniprot", sf.featureGroup);
 180     assertEquals("Pfam", sf.type);
 181     assertEquals(2f, sf.getScore(), 0.001f);
 182   }
 183
 184   public static AlignmentI readAlignmentFile(File f) throws IOException
 185   {
 186     System.out.println("Reading file: " + f);
 187     String ff = f.getPath();
 188     FormatAdapter rf = new FormatAdapter();
 189
 190     AlignmentI al = rf.readFile(ff, FormatAdapter.FILE,
 191             new IdentifyFile().identify(ff, FormatAdapter.FILE));
 192
 193     al.setDataset(null); // creates dataset sequences
 194     assertNotNull("Couldn't read supplied alignment data.", al);
 195     return al;
 196   }
 197
 198   /**
 199    * Test various ways of describing a feature colour scheme
 200    *
 201    * @throws Exception
 202    */
 203   @Test(groups = { "Functional" })
 204   public void testParseGraduatedColourScheme() throws Exception
 205   {
 206     FeaturesFile ff = new FeaturesFile();
 207
 208     // colour by label:
 209     GraduatedColor gc = ff.parseGraduatedColourScheme(
 210             "BETA-TURN-IR\t9a6a94", "label");
 211     assertTrue(gc.isColourByLabel());
 212     assertEquals(Color.white, gc.getMinColor());
 213     assertEquals(Color.black, gc.getMaxColor());
 214     assertTrue(gc.isAutoScale());
 215
 216     // using colour name, rgb, etc:
 217     String spec = "blue|255,0,255|absolute|20.0|95.0|below|66.0";
 218     gc = ff.parseGraduatedColourScheme("BETA-TURN-IR\t" + spec, spec);
 219     assertFalse(gc.isColourByLabel());
 220     assertEquals(Color.blue, gc.getMinColor());
 221     assertEquals(new Color(255, 0, 255), gc.getMaxColor());
 222     assertFalse(gc.isAutoScale());
 223     assertFalse(gc.getTolow());
 224     assertEquals(20.0f, gc.getMin(), 0.001f);
 225     assertEquals(95.0f, gc.getMax(), 0.001f);
 226     assertEquals(AnnotationColourGradient.BELOW_THRESHOLD,
 227             gc.getThreshType());
 228     assertEquals(66.0f, gc.getThresh(), 0.001f);
 229
 230     // inverse gradient high to low:
 231     spec = "blue|255,0,255|95.0|20.0|below|66.0";
 232     gc = ff.parseGraduatedColourScheme("BETA-TURN-IR\t" + spec, spec);
 233     assertTrue(gc.isAutoScale());
 234     assertTrue(gc.getTolow());
 235   }
 236
 237   /**
 238    * Test parsing a features file with GFF formatted content only
 239    *
 240    * @throws Exception
 241    */
 242   @Test(groups = { "Functional" })
 243   public void testParse_pureGff3() throws Exception
 244   {
 245     File f = new File("examples/uniref50.fa");
 246     AlignmentI al = readAlignmentFile(f);
 247     AlignFrame af = new AlignFrame(al, 500, 500);
 248     Map<String, Object> colours = af.getFeatureRenderer()
 249             .getFeatureColours();
 250     // GFF3 uses '=' separator for name/value pairs in colum 9
 251     String gffData = "##gff-version 3\n"
 252             + "FER_CAPAA\tuniprot\tMETAL\t39\t39\t0.0\t.\t.\t"
 253             + "Note=Iron-sulfur (2Fe-2S);Note=another note;evidence=ECO:0000255|PROSITE-ProRule:PRU00465\n"
 254             + "FER1_SOLLC\tuniprot\tPfam\t55\t130\t3.0\t.\t.\tID=$23";
 255     FeaturesFile featuresFile = new FeaturesFile(gffData,
 256             FormatAdapter.PASTE);
 257     assertTrue("Failed to parse features file",
 258             featuresFile.parse(al.getDataset(), colours, true));
 259
 260     // verify feature on FER_CAPAA
 261     SequenceFeature[] sfs = al.getSequenceAt(0).getDatasetSequence()
 262             .getSequenceFeatures();
 263     assertEquals(1, sfs.length);
 264     SequenceFeature sf = sfs[0];
 265     // description parsed from Note attribute
 266     assertEquals("Iron-sulfur (2Fe-2S),another note", sf.description);
 267     assertEquals(39, sf.begin);
 268     assertEquals(39, sf.end);
 269     assertEquals("uniprot", sf.featureGroup);
 270     assertEquals("METAL", sf.type);
 271     assertEquals(
 272             "Note=Iron-sulfur (2Fe-2S);Note=another note;evidence=ECO:0000255|PROSITE-ProRule:PRU00465",
 273             sf.getValue("ATTRIBUTES"));
 274
 275     // verify feature on FER1_SOLLC1
 276     sfs = al.getSequenceAt(2).getDatasetSequence().getSequenceFeatures();
 277     assertEquals(1, sfs.length);
 278     sf = sfs[0];
 279     // ID used for description if available
 280     assertEquals("$23", sf.description);
 281     assertEquals(55, sf.begin);
 282     assertEquals(130, sf.end);
 283     assertEquals("uniprot", sf.featureGroup);
 284     assertEquals("Pfam", sf.type);
 285     assertEquals(3f, sf.getScore(), 0.001f);
 286   }
 287
 288   /**
 289    * Test parsing a features file with Jalview format features (but no colour
 290    * descriptors or startgroup to give the hint not to parse as GFF)
 291    *
 292    * @throws Exception
 293    */
 294   @Test(groups = { "Functional" })
 295   public void testParse_jalviewFeaturesOnly() throws Exception
 296   {
 297     File f = new File("examples/uniref50.fa");
 298     AlignmentI al = readAlignmentFile(f);
 299     AlignFrame af = new AlignFrame(al, 500, 500);
 300     Map<String, Object> colours = af.getFeatureRenderer()
 301             .getFeatureColours();
 302
 303     /*
 304      * one feature on FER_CAPAA and one on sequence 3 (index 2) FER1_SOLLC
 305      */
 306     String featureData = "Iron-sulfur (2Fe-2S)\tFER_CAPAA\t-1\t39\t39\tMETAL\n"
 307             + "Iron-phosphorus (2Fe-P)\tID_NOT_SPECIFIED\t2\t86\t87\tMETALLIC\n";
 308     FeaturesFile featuresFile = new FeaturesFile(featureData,
 309             FormatAdapter.PASTE);
 310     assertTrue("Failed to parse features file",
 311             featuresFile.parse(al.getDataset(), colours, true));
 312
 313     // verify FER_CAPAA feature
 314     SequenceFeature[] sfs = al.getSequenceAt(0).getDatasetSequence()
 315             .getSequenceFeatures();
 316     assertEquals(1, sfs.length);
 317     SequenceFeature sf = sfs[0];
 318     assertEquals("Iron-sulfur (2Fe-2S)", sf.description);
 319     assertEquals(39, sf.begin);
 320     assertEquals(39, sf.end);
 321     assertEquals("METAL", sf.type);
 322
 323     // verify FER1_SOLLC feature
 324     sfs = al.getSequenceAt(2).getDatasetSequence().getSequenceFeatures();
 325     assertEquals(1, sfs.length);
 326     sf = sfs[0];
 327     assertEquals("Iron-phosphorus (2Fe-P)", sf.description);
 328     assertEquals(86, sf.begin);
 329     assertEquals(87, sf.end);
 330     assertEquals("METALLIC", sf.type);
 331   }
 332
 333   private void checkDatasetfromSimpleGff3(AlignmentI dataset)
 334   {
 335     assertEquals("no sequences extracted from GFF3 file", 2,
 336             dataset.getHeight());
 337
 338     SequenceI seq1 = dataset.findName("seq1");
 339     SequenceI seq2 = dataset.findName("seq2");
 340     assertNotNull(seq1);
 341     assertNotNull(seq2);
 342     assertFalse(
 343             "Failed to replace dummy seq1 with real sequence",
 344             seq1 instanceof SequenceDummy
 345                     && ((SequenceDummy) seq1).isDummy());
 346     assertFalse(
 347             "Failed to replace dummy seq2 with real sequence",
 348             seq2 instanceof SequenceDummy
 349                     && ((SequenceDummy) seq2).isDummy());
 350     String placeholderseq = new SequenceDummy("foo").getSequenceAsString();
 351     assertFalse("dummy replacement buggy for seq1",
 352             placeholderseq.equals(seq1.getSequenceAsString()));
 353     assertFalse("dummy replacement buggy for seq2",
 354             placeholderseq.equals(seq2.getSequenceAsString()));
 355     assertNotNull("No features added to seq1", seq1.getSequenceFeatures());
 356     assertEquals("Wrong number of features", 3,
 357             seq1.getSequenceFeatures().length);
 358     assertNull(seq2.getSequenceFeatures());
 359     assertEquals(
 360             "Wrong number of features",
 361             0,
 362             seq2.getSequenceFeatures() == null ? 0 : seq2
 363                     .getSequenceFeatures().length);
 364     assertTrue(
 365             "Expected at least one CDNA/Protein mapping for seq1",
 366             dataset.getCodonFrame(seq1) != null
 367                     && dataset.getCodonFrame(seq1).size() > 0);
 368
 369   }
 370
 371   @Test(groups = { "Functional" })
 372   public void readGff3File() throws IOException
 373   {
 374     FeaturesFile gffreader = new FeaturesFile(true, simpleGffFile,
 375             FormatAdapter.FILE);
 376     Alignment dataset = new Alignment(gffreader.getSeqsAsArray());
 377     gffreader.addProperties(dataset);
 378     checkDatasetfromSimpleGff3(dataset);
 379   }
 380
 381   @Test(groups = { "Functional" })
 382   public void simpleGff3FileClass() throws IOException
 383   {
 384     AlignmentI dataset = new Alignment(new SequenceI[] {});
 385     FeaturesFile ffile = new FeaturesFile(simpleGffFile,
 386             FormatAdapter.FILE);
 387
 388     boolean parseResult = ffile.parse(dataset, null, false, false);
 389     assertTrue("return result should be true", parseResult);
 390     checkDatasetfromSimpleGff3(dataset);
 391   }
 392
 393   @Test(groups = { "Functional" })
 394   public void simpleGff3FileLoader() throws IOException
 395   {
 396     AlignFrame af = new FileLoader(false).LoadFileWaitTillLoaded(
 397             simpleGffFile, FormatAdapter.FILE);
 398     assertTrue(
 399             "Didn't read the alignment into an alignframe from Gff3 File",
 400             af != null);
 401     checkDatasetfromSimpleGff3(af.getViewport().getAlignment());
 402   }
 403
 404   @Test(groups = { "Functional" })
 405   public void simpleGff3RelaxedIdMatching() throws IOException
 406   {
 407     AlignmentI dataset = new Alignment(new SequenceI[] {});
 408     FeaturesFile ffile = new FeaturesFile(simpleGffFile,
 409             FormatAdapter.FILE);
 410
 411     boolean parseResult = ffile.parse(dataset, null, false, true);
 412     assertTrue("return result (relaxedID matching) should be true",
 413             parseResult);
 414     checkDatasetfromSimpleGff3(dataset);
 415   }
 416
 417   @Test(groups = { "Functional" })
 418   public void testPrintJalviewFormat() throws Exception
 419   {
 420     File f = new File("examples/uniref50.fa");
 421     AlignmentI al = readAlignmentFile(f);
 422     AlignFrame af = new AlignFrame(al, 500, 500);
 423     Map<String, Object> colours = af.getFeatureRenderer()
 424             .getFeatureColours();
 425     String features = "METAL\tcc9900\n"
 426             + "GAMMA-TURN\tred|0,255,255|20.0|95.0|below|66.0\n"
 427             + "Pfam\tred\n"
 428             + "STARTGROUP\tuniprot\n"
 429             + "Iron\tFER_CAPAA\t-1\t39\t39\tMETAL\n"
 430             + "Turn\tFER_CAPAA\t-1\t36\t38\tGAMMA-TURN\n"
 431             + "<html>Pfam domain<a href=\"http://pfam.xfam.org/family/PF00111\">Pfam_3_4</a></html>\tFER_CAPAA\t-1\t20\t20\tPfam\n"
 432             + "ENDGROUP\tuniprot\n";
 433     FeaturesFile featuresFile = new FeaturesFile(features,
 434             FormatAdapter.PASTE);
 435     featuresFile.parse(al.getDataset(), colours, false);
 436
 437     /*
 438      * first with no features displayed
 439      */
 440     FeatureRenderer fr = af.alignPanel.getFeatureRenderer();
 441     Map<String, Object> visible = fr
 442             .getDisplayedFeatureCols();
 443     String exported = featuresFile.printJalviewFormat(
 444             al.getSequencesArray(), visible);
 445     String expected = "No Features Visible";
 446     assertEquals(expected, exported);
 447
 448     /*
 449      * set METAL (in uniprot group) and GAMMA-TURN visible, but not Pfam
 450      */
 451     fr.setVisible("METAL");
 452     fr.setVisible("GAMMA-TURN");
 453     visible = fr.getDisplayedFeatureCols();
 454     exported = featuresFile.printJalviewFormat(al.getSequencesArray(),
 455             visible);
 456     expected = "METAL\tcc9900\n"
 457             + "GAMMA-TURN\tff0000|00ffff|20.0|95.0|below|66.0\n"
 458             + "\nSTARTGROUP\tuniprot\n"
 459             + "Iron\tFER_CAPAA\t-1\t39\t39\tMETAL\t0.0\n"
 460             + "Turn\tFER_CAPAA\t-1\t36\t38\tGAMMA-TURN\t0.0\n"
 461             + "ENDGROUP\tuniprot\n";
 462     assertEquals(expected, exported);
 463
 464     /*
 465      * now set Pfam visible
 466      */
 467     fr.setVisible("Pfam");
 468     visible = fr.getDisplayedFeatureCols();
 469     exported = featuresFile.printJalviewFormat(al.getSequencesArray(),
 470             visible);
 471     /*
 472      * note the order of feature types is uncontrolled - derives from
 473      * FeaturesDisplayed.featuresDisplayed which is a HashSet
 474      */
 475     expected = "METAL\tcc9900\n"
 476             + "Pfam\tff0000\n"
 477             + "GAMMA-TURN\tff0000|00ffff|20.0|95.0|below|66.0\n"
 478             + "\nSTARTGROUP\tuniprot\n"
 479             + "Iron\tFER_CAPAA\t-1\t39\t39\tMETAL\t0.0\n"
 480             + "Turn\tFER_CAPAA\t-1\t36\t38\tGAMMA-TURN\t0.0\n"
 481             + "<html>Pfam domain<a href=\"http://pfam.xfam.org/family/PF00111\">Pfam_3_4</a></html>\tFER_CAPAA\t-1\t20\t20\tPfam\t0.0\n"
 482             + "ENDGROUP\tuniprot\n";
 483     assertEquals(expected, exported);
 484   }
 485 }