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.assertSame;
  27 import static org.testng.AssertJUnit.assertTrue;
  28 import static org.testng.internal.junit.ArrayAsserts.assertArrayEquals;
  29
  30 import jalview.api.FeatureColourI;
  31 import jalview.api.FeatureRenderer;
  32 import jalview.datamodel.Alignment;
  33 import jalview.datamodel.AlignmentI;
  34 import jalview.datamodel.SequenceDummy;
  35 import jalview.datamodel.SequenceFeature;
  36 import jalview.datamodel.SequenceI;
  37 import jalview.datamodel.features.FeatureMatcher;
  38 import jalview.datamodel.features.FeatureMatcherI;
  39 import jalview.datamodel.features.FeatureMatcherSet;
  40 import jalview.datamodel.features.FeatureMatcherSetI;
  41 import jalview.datamodel.features.SequenceFeatures;
  42 import jalview.gui.AlignFrame;
  43 import jalview.gui.Desktop;
  44 import jalview.gui.JvOptionPane;
  45 import jalview.schemes.FeatureColour;
  46 import jalview.structure.StructureSelectionManager;
  47 import jalview.util.matcher.Condition;
  48
  49 import java.awt.Color;
  50 import java.io.File;
  51 import java.io.IOException;
  52 import java.util.ArrayList;
  53 import java.util.Arrays;
  54 import java.util.HashMap;
  55 import java.util.Iterator;
  56 import java.util.List;
  57 import java.util.Map;
  58
  59 import org.testng.annotations.AfterClass;
  60 import org.testng.annotations.BeforeClass;
  61 import org.testng.annotations.Test;
  62
  63 public class FeaturesFileTest
  64 {
  65   private static String simpleGffFile = "examples/testdata/simpleGff3.gff";
  66
  67   @AfterClass(alwaysRun = true)
  68   public void tearDownAfterClass()
  69   {
  70     /*
  71      * remove any sequence mappings created so they don't pollute other tests
  72      */
  73     StructureSelectionManager ssm = StructureSelectionManager
  74             .getStructureSelectionManager(Desktop.instance);
  75     ssm.resetAll();
  76   }
  77
  78   @BeforeClass(alwaysRun = true)
  79   public void setUpJvOptionPane()
  80   {
  81     JvOptionPane.setInteractiveMode(false);
  82     JvOptionPane.setMockResponse(JvOptionPane.CANCEL_OPTION);
  83   }
  84
  85   @Test(groups = { "Functional" })
  86   public void testParse() throws Exception
  87   {
  88     File f = new File("examples/uniref50.fa");
  89     AlignmentI al = readAlignmentFile(f);
  90     AlignFrame af = new AlignFrame(al, 500, 500);
  91     Map<String, FeatureColourI> colours = af.getFeatureRenderer()
  92             .getFeatureColours();
  93     FeaturesFile featuresFile = new FeaturesFile(
  94             "examples/exampleFeatures.txt", DataSourceType.FILE);
  95     assertTrue("Test " + "Features file test"
  96             + "\nFailed to parse features file.",
  97             featuresFile.parse(al.getDataset(), colours, true));
  98
  99     /*
 100      * Refetch the colour map from the FeatureRenderer (to confirm it has been
 101      * updated - JAL-1904), and verify (some) feature group colours
 102      */
 103     colours = af.getFeatureRenderer().getFeatureColours();
 104     assertEquals("27 feature group colours not found", 27, colours.size());
 105     assertEquals(colours.get("Cath").getColour(), new Color(0x93b1d1));
 106     assertEquals(colours.get("ASX-MOTIF").getColour(), new Color(0x6addbb));
 107     FeatureColourI kdColour = colours.get("kdHydrophobicity");
 108     assertTrue(kdColour.isGraduatedColour());
 109     assertTrue(kdColour.isAboveThreshold());
 110     assertEquals(-2f, kdColour.getThreshold());
 111
 112     /*
 113      * verify (some) features on sequences
 114      */
 115     List<SequenceFeature> sfs = al.getSequenceAt(0).getDatasetSequence()
 116             .getSequenceFeatures(); // FER_CAPAA
 117     SequenceFeatures.sortFeatures(sfs, true);
 118     assertEquals(8, sfs.size());
 119
 120     /*
 121      * verify (in ascending start position order)
 122      */
 123     SequenceFeature sf = sfs.get(0);
 124     assertEquals("Pfam family%LINK%", sf.description);
 125     assertEquals(0, sf.begin);
 126     assertEquals(0, sf.end);
 127     assertEquals("uniprot", sf.featureGroup);
 128     assertEquals("Pfam", sf.type);
 129     assertEquals(1, sf.links.size());
 130     assertEquals("Pfam family|http://pfam.xfam.org/family/PF00111",
 131             sf.links.get(0));
 132
 133     sf = sfs.get(1);
 134     assertEquals("Ferredoxin_fold Status: True Positive ", sf.description);
 135     assertEquals(3, sf.begin);
 136     assertEquals(93, sf.end);
 137     assertEquals("uniprot", sf.featureGroup);
 138     assertEquals("Cath", sf.type);
 139
 140     sf = sfs.get(2);
 141     assertEquals("Fer2 Status: True Positive Pfam 8_8%LINK%",
 142             sf.description);
 143     assertEquals("Pfam 8_8|http://pfam.xfam.org/family/PF00111",
 144             sf.links.get(0));
 145     assertEquals(8, sf.begin);
 146     assertEquals(83, sf.end);
 147     assertEquals("uniprot", sf.featureGroup);
 148     assertEquals("Pfam", sf.type);
 149
 150     sf = sfs.get(3);
 151     assertEquals("Iron-sulfur (2Fe-2S)", sf.description);
 152     assertEquals(39, sf.begin);
 153     assertEquals(39, sf.end);
 154     assertEquals("uniprot", sf.featureGroup);
 155     assertEquals("METAL", sf.type);
 156
 157     sf = sfs.get(4);
 158     assertEquals("Iron-sulfur (2Fe-2S)", sf.description);
 159     assertEquals(44, sf.begin);
 160     assertEquals(44, sf.end);
 161     assertEquals("uniprot", sf.featureGroup);
 162     assertEquals("METAL", sf.type);
 163
 164     sf = sfs.get(5);
 165     assertEquals("Iron-sulfur (2Fe-2S)", sf.description);
 166     assertEquals(47, sf.begin);
 167     assertEquals(47, sf.end);
 168     assertEquals("uniprot", sf.featureGroup);
 169     assertEquals("METAL", sf.type);
 170
 171     sf = sfs.get(6);
 172     assertEquals("Iron-sulfur (2Fe-2S)", sf.description);
 173     assertEquals(77, sf.begin);
 174     assertEquals(77, sf.end);
 175     assertEquals("uniprot", sf.featureGroup);
 176     assertEquals("METAL", sf.type);
 177
 178     sf = sfs.get(7);
 179     assertEquals(
 180             "High confidence server. Only hits with scores over 0.8 are reported. PHOSPHORYLATION (T) 89_8%LINK%",
 181             sf.description);
 182     assertEquals(
 183             "PHOSPHORYLATION (T) 89_8|http://www.cbs.dtu.dk/cgi-bin/proview/webface-link?seqid=P83527&amp;service=NetPhos-2.0",
 184             sf.links.get(0));
 185     assertEquals(89, sf.begin);
 186     assertEquals(89, sf.end);
 187     assertEquals("netphos", sf.featureGroup);
 188     assertEquals("PHOSPHORYLATION (T)", sf.type);
 189   }
 190
 191   /**
 192    * Test parsing a features file with a mix of Jalview and GFF formatted
 193    * content
 194    *
 195    * @throws Exception
 196    */
 197   @Test(groups = { "Functional" })
 198   public void testParse_mixedJalviewGff() throws Exception
 199   {
 200     File f = new File("examples/uniref50.fa");
 201     AlignmentI al = readAlignmentFile(f);
 202     AlignFrame af = new AlignFrame(al, 500, 500);
 203     Map<String, FeatureColourI> colours = af.getFeatureRenderer()
 204             .getFeatureColours();
 205     // GFF2 uses space as name/value separator in column 9
 206     String gffData = "METAL\tcc9900\n"
 207             + "GFF\n"
 208             + "FER_CAPAA\tuniprot\tMETAL\t44\t45\t4.0\t.\t.\tNote Iron-sulfur; Note 2Fe-2S\n"
 209             + "FER1_SOLLC\tuniprot\tPfam\t55\t130\t2.0\t.\t.";
 210     FeaturesFile featuresFile = new FeaturesFile(gffData,
 211             DataSourceType.PASTE);
 212     assertTrue("Failed to parse features file",
 213             featuresFile.parse(al.getDataset(), colours, true));
 214
 215     // verify colours read or synthesized
 216     colours = af.getFeatureRenderer().getFeatureColours();
 217     assertEquals("1 feature group colours not found", 1, colours.size());
 218     assertEquals(colours.get("METAL").getColour(), new Color(0xcc9900));
 219
 220     // verify feature on FER_CAPAA
 221     List<SequenceFeature> sfs = al.getSequenceAt(0).getDatasetSequence()
 222             .getSequenceFeatures();
 223     assertEquals(1, sfs.size());
 224     SequenceFeature sf = sfs.get(0);
 225     assertEquals("Iron-sulfur,2Fe-2S", sf.description);
 226     assertEquals(44, sf.begin);
 227     assertEquals(45, sf.end);
 228     assertEquals("uniprot", sf.featureGroup);
 229     assertEquals("METAL", sf.type);
 230     assertEquals(4f, sf.getScore(), 0.001f);
 231
 232     // verify feature on FER1_SOLLC
 233     sfs = al.getSequenceAt(2).getDatasetSequence().getSequenceFeatures();
 234     assertEquals(1, sfs.size());
 235     sf = sfs.get(0);
 236     assertEquals("uniprot", sf.description);
 237     assertEquals(55, sf.begin);
 238     assertEquals(130, sf.end);
 239     assertEquals("uniprot", sf.featureGroup);
 240     assertEquals("Pfam", sf.type);
 241     assertEquals(2f, sf.getScore(), 0.001f);
 242   }
 243
 244   public static AlignmentI readAlignmentFile(File f) throws IOException
 245   {
 246     System.out.println("Reading file: " + f);
 247     String ff = f.getPath();
 248     FormatAdapter rf = new FormatAdapter();
 249
 250     AlignmentI al = rf.readFile(ff, DataSourceType.FILE,
 251             new IdentifyFile().identify(ff, DataSourceType.FILE));
 252
 253     al.setDataset(null); // creates dataset sequences
 254     assertNotNull("Couldn't read supplied alignment data.", al);
 255     return al;
 256   }
 257
 258   /**
 259    * Test parsing a features file with GFF formatted content only
 260    *
 261    * @throws Exception
 262    */
 263   @Test(groups = { "Functional" })
 264   public void testParse_pureGff3() throws Exception
 265   {
 266     File f = new File("examples/uniref50.fa");
 267     AlignmentI al = readAlignmentFile(f);
 268     AlignFrame af = new AlignFrame(al, 500, 500);
 269     Map<String, FeatureColourI> colours = af.getFeatureRenderer()
 270             .getFeatureColours();
 271     // GFF3 uses '=' separator for name/value pairs in colum 9
 272     String gffData = "##gff-version 3\n"
 273             + "FER_CAPAA\tuniprot\tMETAL\t39\t39\t0.0\t.\t.\t"
 274             + "Note=Iron-sulfur (2Fe-2S);Note=another note;evidence=ECO:0000255|PROSITE-ProRule:PRU00465\n"
 275             + "FER1_SOLLC\tuniprot\tPfam\t55\t130\t3.0\t.\t.\tID=$23";
 276     FeaturesFile featuresFile = new FeaturesFile(gffData,
 277             DataSourceType.PASTE);
 278     assertTrue("Failed to parse features file",
 279             featuresFile.parse(al.getDataset(), colours, true));
 280
 281     // verify feature on FER_CAPAA
 282     List<SequenceFeature> sfs = al.getSequenceAt(0).getDatasetSequence()
 283             .getSequenceFeatures();
 284     assertEquals(1, sfs.size());
 285     SequenceFeature sf = sfs.get(0);
 286     // description parsed from Note attribute
 287     assertEquals("Iron-sulfur (2Fe-2S),another note", sf.description);
 288     assertEquals(39, sf.begin);
 289     assertEquals(39, sf.end);
 290     assertEquals("uniprot", sf.featureGroup);
 291     assertEquals("METAL", sf.type);
 292     assertEquals(
 293             "Note=Iron-sulfur (2Fe-2S);Note=another note;evidence=ECO:0000255|PROSITE-ProRule:PRU00465",
 294             sf.getValue("ATTRIBUTES"));
 295
 296     // verify feature on FER1_SOLLC1
 297     sfs = al.getSequenceAt(2).getDatasetSequence().getSequenceFeatures();
 298     assertEquals(1, sfs.size());
 299     sf = sfs.get(0);
 300     // ID used for description if available
 301     assertEquals("$23", sf.description);
 302     assertEquals(55, sf.begin);
 303     assertEquals(130, sf.end);
 304     assertEquals("uniprot", sf.featureGroup);
 305     assertEquals("Pfam", sf.type);
 306     assertEquals(3f, sf.getScore(), 0.001f);
 307   }
 308
 309   /**
 310    * Test parsing a features file with Jalview format features (but no colour
 311    * descriptors or startgroup to give the hint not to parse as GFF)
 312    *
 313    * @throws Exception
 314    */
 315   @Test(groups = { "Functional" })
 316   public void testParse_jalviewFeaturesOnly() throws Exception
 317   {
 318     File f = new File("examples/uniref50.fa");
 319     AlignmentI al = readAlignmentFile(f);
 320     AlignFrame af = new AlignFrame(al, 500, 500);
 321     Map<String, FeatureColourI> colours = af.getFeatureRenderer()
 322             .getFeatureColours();
 323
 324     /*
 325      * one feature on FER_CAPAA and one on sequence 3 (index 2) FER1_SOLLC
 326      */
 327     String featureData = "Iron-sulfur (2Fe-2S)\tFER_CAPAA\t-1\t39\t39\tMETAL\n"
 328             + "Iron-phosphorus (2Fe-P)\tID_NOT_SPECIFIED\t2\t86\t87\tMETALLIC\n";
 329     FeaturesFile featuresFile = new FeaturesFile(featureData,
 330             DataSourceType.PASTE);
 331     assertTrue("Failed to parse features file",
 332             featuresFile.parse(al.getDataset(), colours, true));
 333
 334     // verify FER_CAPAA feature
 335     List<SequenceFeature> sfs = al.getSequenceAt(0).getDatasetSequence()
 336             .getSequenceFeatures();
 337     assertEquals(1, sfs.size());
 338     SequenceFeature sf = sfs.get(0);
 339     assertEquals("Iron-sulfur (2Fe-2S)", sf.description);
 340     assertEquals(39, sf.begin);
 341     assertEquals(39, sf.end);
 342     assertEquals("METAL", sf.type);
 343
 344     // verify FER1_SOLLC feature
 345     sfs = al.getSequenceAt(2).getDatasetSequence().getSequenceFeatures();
 346     assertEquals(1, sfs.size());
 347     sf = sfs.get(0);
 348     assertEquals("Iron-phosphorus (2Fe-P)", sf.description);
 349     assertEquals(86, sf.begin);
 350     assertEquals(87, sf.end);
 351     assertEquals("METALLIC", sf.type);
 352   }
 353
 354   private void checkDatasetfromSimpleGff3(AlignmentI dataset)
 355   {
 356     assertEquals("no sequences extracted from GFF3 file", 2,
 357             dataset.getHeight());
 358
 359     SequenceI seq1 = dataset.findName("seq1");
 360     SequenceI seq2 = dataset.findName("seq2");
 361     assertNotNull(seq1);
 362     assertNotNull(seq2);
 363     assertFalse(
 364             "Failed to replace dummy seq1 with real sequence",
 365             seq1 instanceof SequenceDummy
 366                     && ((SequenceDummy) seq1).isDummy());
 367     assertFalse(
 368             "Failed to replace dummy seq2 with real sequence",
 369             seq2 instanceof SequenceDummy
 370                     && ((SequenceDummy) seq2).isDummy());
 371     String placeholderseq = new SequenceDummy("foo").getSequenceAsString();
 372     assertFalse("dummy replacement buggy for seq1",
 373             placeholderseq.equals(seq1.getSequenceAsString()));
 374     assertFalse("dummy replacement buggy for seq2",
 375             placeholderseq.equals(seq2.getSequenceAsString()));
 376     assertNotNull("No features added to seq1", seq1.getSequenceFeatures());
 377     assertEquals("Wrong number of features", 3, seq1.getSequenceFeatures()
 378             .size());
 379     assertTrue(seq2.getSequenceFeatures().isEmpty());
 380     assertEquals(
 381             "Wrong number of features",
 382             0,
 383             seq2.getSequenceFeatures() == null ? 0 : seq2
 384                     .getSequenceFeatures().size());
 385     assertTrue(
 386             "Expected at least one CDNA/Protein mapping for seq1",
 387             dataset.getCodonFrame(seq1) != null
 388                     && dataset.getCodonFrame(seq1).size() > 0);
 389
 390   }
 391
 392   @Test(groups = { "Functional" })
 393   public void readGff3File() throws IOException
 394   {
 395     FeaturesFile gffreader = new FeaturesFile(true, simpleGffFile,
 396             DataSourceType.FILE);
 397     Alignment dataset = new Alignment(gffreader.getSeqsAsArray());
 398     gffreader.addProperties(dataset);
 399     checkDatasetfromSimpleGff3(dataset);
 400   }
 401
 402   @Test(groups = { "Functional" })
 403   public void simpleGff3FileClass() throws IOException
 404   {
 405     AlignmentI dataset = new Alignment(new SequenceI[] {});
 406     FeaturesFile ffile = new FeaturesFile(simpleGffFile,
 407             DataSourceType.FILE);
 408
 409     boolean parseResult = ffile.parse(dataset, null, false, false);
 410     assertTrue("return result should be true", parseResult);
 411     checkDatasetfromSimpleGff3(dataset);
 412   }
 413
 414   @Test(groups = { "Functional" })
 415   public void simpleGff3FileLoader() throws IOException
 416   {
 417     AlignFrame af = new FileLoader(false).LoadFileWaitTillLoaded(
 418             simpleGffFile, DataSourceType.FILE);
 419     assertTrue(
 420             "Didn't read the alignment into an alignframe from Gff3 File",
 421             af != null);
 422     checkDatasetfromSimpleGff3(af.getViewport().getAlignment());
 423   }
 424
 425   @Test(groups = { "Functional" })
 426   public void simpleGff3RelaxedIdMatching() throws IOException
 427   {
 428     AlignmentI dataset = new Alignment(new SequenceI[] {});
 429     FeaturesFile ffile = new FeaturesFile(simpleGffFile,
 430             DataSourceType.FILE);
 431
 432     boolean parseResult = ffile.parse(dataset, null, false, true);
 433     assertTrue("return result (relaxedID matching) should be true",
 434             parseResult);
 435     checkDatasetfromSimpleGff3(dataset);
 436   }
 437
 438   @Test(groups = { "Functional" })
 439   public void testPrintJalviewFormat() throws Exception
 440   {
 441     File f = new File("examples/uniref50.fa");
 442     AlignmentI al = readAlignmentFile(f);
 443     AlignFrame af = new AlignFrame(al, 500, 500);
 444     Map<String, FeatureColourI> colours = af.getFeatureRenderer()
 445             .getFeatureColours();
 446     String features = "METAL\tcc9900\n"
 447             + "GAMMA-TURN\tred|0,255,255|20.0|95.0|below|66.0\n"
 448             + "Pfam\tred\n"
 449             + "STARTGROUP\tuniprot\n"
 450             + "Cath\tFER_CAPAA\t-1\t0\t0\tDomain\n" // non-positional feature
 451             + "Iron\tFER_CAPAA\t-1\t39\t39\tMETAL\n"
 452             + "Turn\tFER_CAPAA\t-1\t36\t38\tGAMMA-TURN\n"
 453             + "<html>Pfam domain<a href=\"http://pfam.xfam.org/family/PF00111\">Pfam_3_4</a></html>\tFER_CAPAA\t-1\t20\t20\tPfam\n"
 454             + "ENDGROUP\tuniprot\n";
 455     FeaturesFile featuresFile = new FeaturesFile(features,
 456             DataSourceType.PASTE);
 457     featuresFile.parse(al.getDataset(), colours, false);
 458
 459     /*
 460      * add positional and non-positional features with null and
 461      * empty feature group to check handled correctly
 462      */
 463     SequenceI seq = al.getSequenceAt(1); // FER_CAPAN
 464     seq.addSequenceFeature(new SequenceFeature("Pfam", "desc1", 0, 0, 1.3f,
 465             null));
 466     seq.addSequenceFeature(new SequenceFeature("Pfam", "desc2", 4, 9,
 467             Float.NaN, null));
 468     seq = al.getSequenceAt(2); // FER1_SOLLC
 469     seq.addSequenceFeature(new SequenceFeature("Pfam", "desc3", 0, 0,
 470             Float.NaN, ""));
 471     seq.addSequenceFeature(new SequenceFeature("Pfam", "desc4", 5, 8,
 472             -2.6f, ""));
 473
 474     /*
 475      * first with no features displayed, exclude non-positional features
 476      */
 477     FeatureRenderer fr = af.alignPanel.getFeatureRenderer();
 478     Map<String, FeatureColourI> visible = fr.getDisplayedFeatureCols();
 479     List<String> visibleGroups = new ArrayList<>(
 480             Arrays.asList(new String[] {}));
 481     String exported = featuresFile.printJalviewFormat(
 482             al.getSequencesArray(), visible, null, visibleGroups, false);
 483     String expected = "No Features Visible";
 484     assertEquals(expected, exported);
 485
 486     /*
 487      * include non-positional features
 488      */
 489     visibleGroups.add("uniprot");
 490     exported = featuresFile.printJalviewFormat(al.getSequencesArray(),
 491             visible, null, visibleGroups, true);
 492     expected = "Cath\tFER_CAPAA\t-1\t0\t0\tDomain\t0.0\n"
 493             + "desc1\tFER_CAPAN\t-1\t0\t0\tPfam\t1.3\n"
 494             + "desc3\tFER1_SOLLC\t-1\t0\t0\tPfam\n" // NaN is not output
 495             + "\nSTARTGROUP\tuniprot\nENDGROUP\tuniprot\n";
 496     assertEquals(expected, exported);
 497
 498     /*
 499      * set METAL (in uniprot group) and GAMMA-TURN visible, but not Pfam
 500      */
 501     fr.setVisible("METAL");
 502     fr.setVisible("GAMMA-TURN");
 503     visible = fr.getDisplayedFeatureCols();
 504     exported = featuresFile.printJalviewFormat(al.getSequencesArray(),
 505             visible, null, visibleGroups, false);
 506     expected = "METAL\tcc9900\n"
 507             + "GAMMA-TURN\tscore|ff0000|00ffff|noValueMin|20.0|95.0|below|66.0\n"
 508             + "\nSTARTGROUP\tuniprot\n"
 509             + "Turn\tFER_CAPAA\t-1\t36\t38\tGAMMA-TURN\t0.0\n"
 510             + "Iron\tFER_CAPAA\t-1\t39\t39\tMETAL\t0.0\n"
 511             + "ENDGROUP\tuniprot\n";
 512     assertEquals(expected, exported);
 513
 514     /*
 515      * now set Pfam visible
 516      */
 517     fr.setVisible("Pfam");
 518     visible = fr.getDisplayedFeatureCols();
 519     exported = featuresFile.printJalviewFormat(al.getSequencesArray(),
 520             visible, null, visibleGroups, false);
 521     /*
 522      * features are output within group, ordered by sequence and by type
 523      */
 524     expected = "METAL\tcc9900\n"
 525             + "Pfam\tff0000\n"
 526             + "GAMMA-TURN\tscore|ff0000|00ffff|noValueMin|20.0|95.0|below|66.0\n"
 527             + "\nSTARTGROUP\tuniprot\n"
 528             + "Turn\tFER_CAPAA\t-1\t36\t38\tGAMMA-TURN\t0.0\n"
 529             + "Iron\tFER_CAPAA\t-1\t39\t39\tMETAL\t0.0\n"
 530             + "<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"
 531             + "ENDGROUP\tuniprot\n"
 532             // null / empty group features output after features in named
 533             // groups:
 534             + "desc2\tFER_CAPAN\t-1\t4\t9\tPfam\n"
 535             + "desc4\tFER1_SOLLC\t-1\t5\t8\tPfam\t-2.6\n";
 536     assertEquals(expected, exported);
 537   }
 538
 539   @Test(groups = { "Functional" })
 540   public void testPrintGffFormat() throws Exception
 541   {
 542     File f = new File("examples/uniref50.fa");
 543     AlignmentI al = readAlignmentFile(f);
 544     AlignFrame af = new AlignFrame(al, 500, 500);
 545
 546     /*
 547      * no features
 548      */
 549     FeaturesFile featuresFile = new FeaturesFile();
 550     FeatureRenderer fr = af.alignPanel.getFeatureRenderer();
 551     Map<String, FeatureColourI> visible = new HashMap<>();
 552     List<String> visibleGroups = new ArrayList<>(
 553             Arrays.asList(new String[] {}));
 554     String exported = featuresFile.printGffFormat(al.getSequencesArray(),
 555             visible, visibleGroups, false);
 556     String gffHeader = "##gff-version 2\n";
 557     assertEquals(gffHeader, exported);
 558     exported = featuresFile.printGffFormat(al.getSequencesArray(), visible,
 559             visibleGroups, true);
 560     assertEquals(gffHeader, exported);
 561
 562     /*
 563      * add some features
 564      */
 565     al.getSequenceAt(0).addSequenceFeature(
 566             new SequenceFeature("Domain", "Cath", 0, 0, 0f, "Uniprot"));
 567     al.getSequenceAt(0).addSequenceFeature(
 568             new SequenceFeature("METAL", "Cath", 39, 39, 1.2f, null));
 569     al.getSequenceAt(1)
 570             .addSequenceFeature(
 571                     new SequenceFeature("GAMMA-TURN", "Turn", 36, 38, 2.1f,
 572                             "s3dm"));
 573     SequenceFeature sf = new SequenceFeature("Pfam", "", 20, 20, 0f,
 574             "Uniprot");
 575     sf.setAttributes("x=y;black=white");
 576     sf.setStrand("+");
 577     sf.setPhase("2");
 578     al.getSequenceAt(1).addSequenceFeature(sf);
 579
 580     /*
 581      * with no features displayed, exclude non-positional features
 582      */
 583     exported = featuresFile.printGffFormat(al.getSequencesArray(), visible,
 584             visibleGroups, false);
 585     assertEquals(gffHeader, exported);
 586
 587     /*
 588      * include non-positional features
 589      */
 590     visibleGroups.add("Uniprot");
 591     exported = featuresFile.printGffFormat(al.getSequencesArray(), visible,
 592             visibleGroups, true);
 593     String expected = gffHeader
 594             + "FER_CAPAA\tUniprot\tDomain\t0\t0\t0.0\t.\t.\n";
 595     assertEquals(expected, exported);
 596
 597     /*
 598      * set METAL (in uniprot group) and GAMMA-TURN visible, but not Pfam
 599      * only Uniprot group visible here...
 600      */
 601     fr.setVisible("METAL");
 602     fr.setVisible("GAMMA-TURN");
 603     visible = fr.getDisplayedFeatureCols();
 604     exported = featuresFile.printGffFormat(al.getSequencesArray(), visible,
 605             visibleGroups, false);
 606     // METAL feature has null group: description used for column 2
 607     expected = gffHeader + "FER_CAPAA\tCath\tMETAL\t39\t39\t1.2\t.\t.\n";
 608     assertEquals(expected, exported);
 609
 610     /*
 611      * set s3dm group visible
 612      */
 613     visibleGroups.add("s3dm");
 614     exported = featuresFile.printGffFormat(al.getSequencesArray(), visible,
 615             visibleGroups, false);
 616     // METAL feature has null group: description used for column 2
 617     expected = gffHeader + "FER_CAPAA\tCath\tMETAL\t39\t39\t1.2\t.\t.\n"
 618             + "FER_CAPAN\ts3dm\tGAMMA-TURN\t36\t38\t2.1\t.\t.\n";
 619     assertEquals(expected, exported);
 620
 621     /*
 622      * now set Pfam visible
 623      */
 624     fr.setVisible("Pfam");
 625     visible = fr.getDisplayedFeatureCols();
 626     exported = featuresFile.printGffFormat(al.getSequencesArray(), visible,
 627             visibleGroups, false);
 628     // Pfam feature columns include strand(+), phase(2), attributes
 629     expected = gffHeader
 630             + "FER_CAPAA\tCath\tMETAL\t39\t39\t1.2\t.\t.\n"
 631             + "FER_CAPAN\ts3dm\tGAMMA-TURN\t36\t38\t2.1\t.\t.\n"
 632             + "FER_CAPAN\tUniprot\tPfam\t20\t20\t0.0\t+\t2\tx=y;black=white\n";
 633     assertEquals(expected, exported);
 634   }
 635
 636   /**
 637    * Test for parsing of feature filters as represented in a Jalview features
 638    * file
 639    *
 640    * @throws Exception
 641    */
 642   @Test(groups = { "Functional" })
 643   public void testParseFilters() throws Exception
 644   {
 645     Map<String, FeatureMatcherSetI> filters = new HashMap<>();
 646     String text = "sequence_variant\tCSQ:PolyPhen NotContains 'damaging'\n"
 647             + "missense_variant\t(label contains foobar) and (Score lt 1.3)";
 648     FeaturesFile featuresFile = new FeaturesFile(text,
 649             DataSourceType.PASTE);
 650     featuresFile.parseFilters(filters);
 651     assertEquals(filters.size(), 2);
 652
 653     FeatureMatcherSetI fm = filters.get("sequence_variant");
 654     assertNotNull(fm);
 655     Iterator<FeatureMatcherI> matchers = fm.getMatchers().iterator();
 656     FeatureMatcherI matcher = matchers.next();
 657     assertFalse(matchers.hasNext());
 658     String[] attributes = matcher.getAttribute();
 659     assertArrayEquals(attributes, new String[] { "CSQ", "PolyPhen" });
 660     assertSame(matcher.getMatcher().getCondition(), Condition.NotContains);
 661     assertEquals(matcher.getMatcher().getPattern(), "damaging");
 662
 663     fm = filters.get("missense_variant");
 664     assertNotNull(fm);
 665     matchers = fm.getMatchers().iterator();
 666     matcher = matchers.next();
 667     assertTrue(matcher.isByLabel());
 668     assertSame(matcher.getMatcher().getCondition(), Condition.Contains);
 669     assertEquals(matcher.getMatcher().getPattern(), "foobar");
 670     matcher = matchers.next();
 671     assertTrue(matcher.isByScore());
 672     assertSame(matcher.getMatcher().getCondition(), Condition.LT);
 673     assertEquals(matcher.getMatcher().getPattern(), "1.3");
 674     assertEquals(matcher.getMatcher().getFloatValue(), 1.3f);
 675
 676     assertFalse(matchers.hasNext());
 677   }
 678
 679   @Test(groups = { "Functional" })
 680   public void testOutputFeatureFilters()
 681   {
 682     FeaturesFile ff = new FeaturesFile();
 683     StringBuilder sb = new StringBuilder();
 684     Map<String, FeatureColourI> visible = new HashMap<>();
 685     visible.put("pfam", new FeatureColour(Color.red));
 686     Map<String, FeatureMatcherSetI> featureFilters = new HashMap<>();
 687
 688     // with no filters, nothing is output
 689     ff.outputFeatureFilters(sb, visible, featureFilters);
 690     assertEquals("", sb.toString());
 691
 692     // with filter for not visible features only, nothing is output
 693     FeatureMatcherSet filter = new FeatureMatcherSet();
 694     filter.and(FeatureMatcher.byLabel(Condition.Present, null));
 695     featureFilters.put("foobar", filter);
 696     ff.outputFeatureFilters(sb, visible, featureFilters);
 697     assertEquals("", sb.toString());
 698
 699     // with filters for visible feature types
 700     FeatureMatcherSet filter2 = new FeatureMatcherSet();
 701     filter2.and(FeatureMatcher.byAttribute(Condition.Present, null, "CSQ",
 702             "PolyPhen"));
 703     filter2.and(FeatureMatcher.byScore(Condition.LE, "-2.4"));
 704     featureFilters.put("pfam", filter2);
 705     visible.put("foobar", new FeatureColour(Color.blue));
 706     ff.outputFeatureFilters(sb, visible, featureFilters);
 707     String expected = "\nSTARTFILTERS\nfoobar\tLabel Present\npfam\t(CSQ:PolyPhen Present) AND (Score LE -2.4)\nENDFILTERS\n\n";
 708     assertEquals(expected, sb.toString());
 709   }
 710 }