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(8, sfs.length);
  80     SequenceFeature sf = sfs[0];
  81     assertEquals("Pfam family%LINK%", sf.description);
  82     assertEquals(0, sf.begin);
  83     assertEquals(0, sf.end);
  84     assertEquals("uniprot", sf.featureGroup);
  85     assertEquals("Pfam", sf.type);
  86     assertEquals(1, sf.links.size());
  87     assertEquals("Pfam family|http://pfam.xfam.org/family/PF00111",
  88             sf.links.get(0));
  89
  90     sf = sfs[1];
  91     assertEquals("Iron-sulfur (2Fe-2S)", sf.description);
  92     assertEquals(39, sf.begin);
  93     assertEquals(39, sf.end);
  94     assertEquals("uniprot", sf.featureGroup);
  95     assertEquals("METAL", sf.type);
  96     sf = sfs[2];
  97     assertEquals("Iron-sulfur (2Fe-2S)", sf.description);
  98     assertEquals(44, sf.begin);
  99     assertEquals(44, sf.end);
 100     assertEquals("uniprot", sf.featureGroup);
 101     assertEquals("METAL", sf.type);
 102     sf = sfs[3];
 103     assertEquals("Iron-sulfur (2Fe-2S)", sf.description);
 104     assertEquals(47, sf.begin);
 105     assertEquals(47, sf.end);
 106     assertEquals("uniprot", sf.featureGroup);
 107     assertEquals("METAL", sf.type);
 108     sf = sfs[4];
 109     assertEquals("Iron-sulfur (2Fe-2S)", sf.description);
 110     assertEquals(77, sf.begin);
 111     assertEquals(77, sf.end);
 112     assertEquals("uniprot", sf.featureGroup);
 113     assertEquals("METAL", sf.type);
 114     sf = sfs[5];
 115     assertEquals("Fer2 Status: True Positive Pfam 8_8%LINK%",
 116             sf.description);
 117     assertEquals("Pfam 8_8|http://pfam.xfam.org/family/PF00111",
 118             sf.links.get(0));
 119     assertEquals(8, sf.begin);
 120     assertEquals(83, sf.end);
 121     assertEquals("uniprot", sf.featureGroup);
 122     assertEquals("Pfam", sf.type);
 123     sf = sfs[6];
 124     assertEquals("Ferredoxin_fold Status: True Positive ", sf.description);
 125     assertEquals(3, sf.begin);
 126     assertEquals(93, sf.end);
 127     assertEquals("uniprot", sf.featureGroup);
 128     assertEquals("Cath", sf.type);
 129     sf = sfs[7];
 130     assertEquals(
 131             "High confidence server. Only hits with scores over 0.8 are reported. PHOSPHORYLATION (T) 89_8%LINK%",
 132             sf.description);
 133     assertEquals(
 134             "PHOSPHORYLATION (T) 89_8|http://www.cbs.dtu.dk/cgi-bin/proview/webface-link?seqid=P83527&amp;service=NetPhos-2.0",
 135             sf.links.get(0));
 136     assertEquals(89, sf.begin);
 137     assertEquals(89, sf.end);
 138     assertEquals("netphos", sf.featureGroup);
 139     assertEquals("PHOSPHORYLATION (T)", sf.type);
 140   }
 141
 142   /**
 143    * Test parsing a features file with a mix of Jalview and GFF formatted
 144    * content
 145    *
 146    * @throws Exception
 147    */
 148   @Test(groups = { "Functional" })
 149   public void testParse_mixedJalviewGff() throws Exception
 150   {
 151     File f = new File("examples/uniref50.fa");
 152     AlignmentI al = readAlignmentFile(f);
 153     AlignFrame af = new AlignFrame(al, 500, 500);
 154     Map<String, Object> colours = af.getFeatureRenderer()
 155             .getFeatureColours();
 156     // GFF2 uses space as name/value separator in column 9
 157     String gffData = "METAL\tcc9900\n" + "GFF\n"
 158             + "FER_CAPAA\tuniprot\tMETAL\t44\t45\t4.0\t.\t.\tNote Iron-sulfur; Note 2Fe-2S\n"
 159             + "FER1_SOLLC\tuniprot\tPfam\t55\t130\t2.0\t.\t.";
 160     FeaturesFile featuresFile = new FeaturesFile(gffData,
 161             FormatAdapter.PASTE);
 162     assertTrue("Failed to parse features file",
 163             featuresFile.parse(al.getDataset(), colours, true));
 164
 165     // verify colours read or synthesized
 166     colours = af.getFeatureRenderer().getFeatureColours();
 167     assertEquals("1 feature group colours not found", 1, colours.size());
 168     assertEquals(colours.get("METAL"), new Color(0xcc9900));
 169
 170     // verify feature on FER_CAPAA
 171     SequenceFeature[] sfs = al.getSequenceAt(0).getDatasetSequence()
 172             .getSequenceFeatures();
 173     assertEquals(1, sfs.length);
 174     SequenceFeature sf = sfs[0];
 175     assertEquals("Iron-sulfur,2Fe-2S", sf.description);
 176     assertEquals(44, sf.begin);
 177     assertEquals(45, sf.end);
 178     assertEquals("uniprot", sf.featureGroup);
 179     assertEquals("METAL", sf.type);
 180     assertEquals(4f, sf.getScore(), 0.001f);
 181
 182     // verify feature on FER1_SOLLC
 183     sfs = al.getSequenceAt(2).getDatasetSequence().getSequenceFeatures();
 184     assertEquals(1, sfs.length);
 185     sf = sfs[0];
 186     assertEquals("uniprot", sf.description);
 187     assertEquals(55, sf.begin);
 188     assertEquals(130, sf.end);
 189     assertEquals("uniprot", sf.featureGroup);
 190     assertEquals("Pfam", sf.type);
 191     assertEquals(2f, sf.getScore(), 0.001f);
 192   }
 193
 194   public static AlignmentI readAlignmentFile(File f) throws IOException
 195   {
 196     System.out.println("Reading file: " + f);
 197     String ff = f.getPath();
 198     FormatAdapter rf = new FormatAdapter();
 199
 200     AlignmentI al = rf.readFile(ff, FormatAdapter.FILE,
 201             new IdentifyFile().identify(ff, FormatAdapter.FILE));
 202
 203     al.setDataset(null); // creates dataset sequences
 204     assertNotNull("Couldn't read supplied alignment data.", al);
 205     return al;
 206   }
 207
 208   /**
 209    * Test various ways of describing a feature colour scheme
 210    *
 211    * @throws Exception
 212    */
 213   @Test(groups = { "Functional" })
 214   public void testParseGraduatedColourScheme() throws Exception
 215   {
 216     FeaturesFile ff = new FeaturesFile();
 217
 218     // colour by label:
 219     GraduatedColor gc = ff.parseGraduatedColourScheme(
 220             "BETA-TURN-IR\t9a6a94", "label");
 221     assertTrue(gc.isColourByLabel());
 222     assertEquals(Color.white, gc.getMinColor());
 223     assertEquals(Color.black, gc.getMaxColor());
 224     assertTrue(gc.isAutoScale());
 225
 226     // using colour name, rgb, etc:
 227     String spec = "blue|255,0,255|absolute|20.0|95.0|below|66.0";
 228     gc = ff.parseGraduatedColourScheme("BETA-TURN-IR\t" + spec, spec);
 229     assertFalse(gc.isColourByLabel());
 230     assertEquals(Color.blue, gc.getMinColor());
 231     assertEquals(new Color(255, 0, 255), gc.getMaxColor());
 232     assertFalse(gc.isAutoScale());
 233     assertFalse(gc.getTolow());
 234     assertEquals(20.0f, gc.getMin(), 0.001f);
 235     assertEquals(95.0f, gc.getMax(), 0.001f);
 236     assertEquals(AnnotationColourGradient.BELOW_THRESHOLD,
 237             gc.getThreshType());
 238     assertEquals(66.0f, gc.getThresh(), 0.001f);
 239
 240     // inverse gradient high to low:
 241     spec = "blue|255,0,255|95.0|20.0|below|66.0";
 242     gc = ff.parseGraduatedColourScheme("BETA-TURN-IR\t" + spec, spec);
 243     assertTrue(gc.isAutoScale());
 244     assertTrue(gc.getTolow());
 245   }
 246
 247   /**
 248    * Test parsing a features file with GFF formatted content only
 249    *
 250    * @throws Exception
 251    */
 252   @Test(groups = { "Functional" })
 253   public void testParse_pureGff3() throws Exception
 254   {
 255     File f = new File("examples/uniref50.fa");
 256     AlignmentI al = readAlignmentFile(f);
 257     AlignFrame af = new AlignFrame(al, 500, 500);
 258     Map<String, Object> colours = af.getFeatureRenderer()
 259             .getFeatureColours();
 260     // GFF3 uses '=' separator for name/value pairs in colum 9
 261     String gffData = "##gff-version 3\n"
 262             + "FER_CAPAA\tuniprot\tMETAL\t39\t39\t0.0\t.\t.\t"
 263             + "Note=Iron-sulfur (2Fe-2S);Note=another note;evidence=ECO:0000255|PROSITE-ProRule:PRU00465\n"
 264             + "FER1_SOLLC\tuniprot\tPfam\t55\t130\t3.0\t.\t.\tID=$23";
 265     FeaturesFile featuresFile = new FeaturesFile(gffData,
 266             FormatAdapter.PASTE);
 267     assertTrue("Failed to parse features file",
 268             featuresFile.parse(al.getDataset(), colours, true));
 269
 270     // verify feature on FER_CAPAA
 271     SequenceFeature[] sfs = al.getSequenceAt(0).getDatasetSequence()
 272             .getSequenceFeatures();
 273     assertEquals(1, sfs.length);
 274     SequenceFeature sf = sfs[0];
 275     // description parsed from Note attribute
 276     assertEquals("Iron-sulfur (2Fe-2S),another note", sf.description);
 277     assertEquals(39, sf.begin);
 278     assertEquals(39, sf.end);
 279     assertEquals("uniprot", sf.featureGroup);
 280     assertEquals("METAL", sf.type);
 281     assertEquals(
 282             "Note=Iron-sulfur (2Fe-2S);Note=another note;evidence=ECO:0000255|PROSITE-ProRule:PRU00465",
 283             sf.getValue("ATTRIBUTES"));
 284
 285     // verify feature on FER1_SOLLC1
 286     sfs = al.getSequenceAt(2).getDatasetSequence().getSequenceFeatures();
 287     assertEquals(1, sfs.length);
 288     sf = sfs[0];
 289     // ID used for description if available
 290     assertEquals("$23", sf.description);
 291     assertEquals(55, sf.begin);
 292     assertEquals(130, sf.end);
 293     assertEquals("uniprot", sf.featureGroup);
 294     assertEquals("Pfam", sf.type);
 295     assertEquals(3f, sf.getScore(), 0.001f);
 296   }
 297
 298   /**
 299    * Test parsing a features file with Jalview format features (but no colour
 300    * descriptors or startgroup to give the hint not to parse as GFF)
 301    *
 302    * @throws Exception
 303    */
 304   @Test(groups = { "Functional" })
 305   public void testParse_jalviewFeaturesOnly() throws Exception
 306   {
 307     File f = new File("examples/uniref50.fa");
 308     AlignmentI al = readAlignmentFile(f);
 309     AlignFrame af = new AlignFrame(al, 500, 500);
 310     Map<String, Object> colours = af.getFeatureRenderer()
 311             .getFeatureColours();
 312
 313     /*
 314      * one feature on FER_CAPAA and one on sequence 3 (index 2) FER1_SOLLC
 315      */
 316     String featureData = "Iron-sulfur (2Fe-2S)\tFER_CAPAA\t-1\t39\t39\tMETAL\n"
 317             + "Iron-phosphorus (2Fe-P)\tID_NOT_SPECIFIED\t2\t86\t87\tMETALLIC\n";
 318     FeaturesFile featuresFile = new FeaturesFile(featureData,
 319             FormatAdapter.PASTE);
 320     assertTrue("Failed to parse features file",
 321             featuresFile.parse(al.getDataset(), colours, true));
 322
 323     // verify FER_CAPAA feature
 324     SequenceFeature[] sfs = al.getSequenceAt(0).getDatasetSequence()
 325             .getSequenceFeatures();
 326     assertEquals(1, sfs.length);
 327     SequenceFeature sf = sfs[0];
 328     assertEquals("Iron-sulfur (2Fe-2S)", sf.description);
 329     assertEquals(39, sf.begin);
 330     assertEquals(39, sf.end);
 331     assertEquals("METAL", sf.type);
 332
 333     // verify FER1_SOLLC feature
 334     sfs = al.getSequenceAt(2).getDatasetSequence().getSequenceFeatures();
 335     assertEquals(1, sfs.length);
 336     sf = sfs[0];
 337     assertEquals("Iron-phosphorus (2Fe-P)", sf.description);
 338     assertEquals(86, sf.begin);
 339     assertEquals(87, sf.end);
 340     assertEquals("METALLIC", sf.type);
 341   }
 342
 343   private void checkDatasetfromSimpleGff3(AlignmentI dataset)
 344   {
 345     assertEquals("no sequences extracted from GFF3 file", 2,
 346             dataset.getHeight());
 347
 348     SequenceI seq1 = dataset.findName("seq1");
 349     SequenceI seq2 = dataset.findName("seq2");
 350     assertNotNull(seq1);
 351     assertNotNull(seq2);
 352     assertFalse(
 353             "Failed to replace dummy seq1 with real sequence",
 354             seq1 instanceof SequenceDummy
 355                     && ((SequenceDummy) seq1).isDummy());
 356     assertFalse(
 357             "Failed to replace dummy seq2 with real sequence",
 358             seq2 instanceof SequenceDummy
 359                     && ((SequenceDummy) seq2).isDummy());
 360     String placeholderseq = new SequenceDummy("foo").getSequenceAsString();
 361     assertFalse("dummy replacement buggy for seq1",
 362             placeholderseq.equals(seq1.getSequenceAsString()));
 363     assertFalse("dummy replacement buggy for seq2",
 364             placeholderseq.equals(seq2.getSequenceAsString()));
 365     assertNotNull("No features added to seq1", seq1.getSequenceFeatures());
 366     assertEquals("Wrong number of features", 3,
 367             seq1.getSequenceFeatures().length);
 368     assertNull(seq2.getSequenceFeatures());
 369     assertEquals(
 370             "Wrong number of features",
 371             0,
 372             seq2.getSequenceFeatures() == null ? 0 : seq2
 373                     .getSequenceFeatures().length);
 374     assertTrue(
 375             "Expected at least one CDNA/Protein mapping for seq1",
 376             dataset.getCodonFrame(seq1) != null
 377                     && dataset.getCodonFrame(seq1).size() > 0);
 378
 379   }
 380
 381   @Test(groups = { "Functional" })
 382   public void readGff3File() throws IOException
 383   {
 384     FeaturesFile gffreader = new FeaturesFile(true, simpleGffFile,
 385             FormatAdapter.FILE);
 386     Alignment dataset = new Alignment(gffreader.getSeqsAsArray());
 387     gffreader.addProperties(dataset);
 388     checkDatasetfromSimpleGff3(dataset);
 389   }
 390
 391   @Test(groups = { "Functional" })
 392   public void simpleGff3FileClass() throws IOException
 393   {
 394     AlignmentI dataset = new Alignment(new SequenceI[] {});
 395     FeaturesFile ffile = new FeaturesFile(simpleGffFile,
 396             FormatAdapter.FILE);
 397
 398     boolean parseResult = ffile.parse(dataset, null, false, false);
 399     assertTrue("return result should be true", parseResult);
 400     checkDatasetfromSimpleGff3(dataset);
 401   }
 402
 403   @Test(groups = { "Functional" })
 404   public void simpleGff3FileLoader() throws IOException
 405   {
 406     AlignFrame af = new FileLoader(false).LoadFileWaitTillLoaded(
 407             simpleGffFile, FormatAdapter.FILE);
 408     assertTrue(
 409             "Didn't read the alignment into an alignframe from Gff3 File",
 410             af != null);
 411     checkDatasetfromSimpleGff3(af.getViewport().getAlignment());
 412   }
 413
 414   @Test(groups = { "Functional" })
 415   public void simpleGff3RelaxedIdMatching() throws IOException
 416   {
 417     AlignmentI dataset = new Alignment(new SequenceI[] {});
 418     FeaturesFile ffile = new FeaturesFile(simpleGffFile,
 419             FormatAdapter.FILE);
 420
 421     boolean parseResult = ffile.parse(dataset, null, false, true);
 422     assertTrue("return result (relaxedID matching) should be true",
 423             parseResult);
 424     checkDatasetfromSimpleGff3(dataset);
 425   }
 426
 427   @Test(groups = { "Functional" })
 428   public void testPrintJalviewFormat() throws Exception
 429   {
 430     File f = new File("examples/uniref50.fa");
 431     AlignmentI al = readAlignmentFile(f);
 432     AlignFrame af = new AlignFrame(al, 500, 500);
 433     Map<String, Object> colours = af.getFeatureRenderer()
 434             .getFeatureColours();
 435     String features = "METAL\tcc9900\n"
 436             + "GAMMA-TURN\tred|0,255,255|20.0|95.0|below|66.0\n"
 437             + "Pfam\tred\n"
 438             + "STARTGROUP\tuniprot\n"
 439             + "Iron\tFER_CAPAA\t-1\t39\t39\tMETAL\n"
 440             + "Turn\tFER_CAPAA\t-1\t36\t38\tGAMMA-TURN\n"
 441             + "<html>Pfam domain<a href=\"http://pfam.xfam.org/family/PF00111\">Pfam_3_4</a></html>\tFER_CAPAA\t-1\t20\t20\tPfam\n"
 442             + "ENDGROUP\tuniprot\n";
 443     FeaturesFile featuresFile = new FeaturesFile(features,
 444             FormatAdapter.PASTE);
 445     featuresFile.parse(al.getDataset(), colours, false);
 446
 447     /*
 448      * first with no features displayed
 449      */
 450     FeatureRenderer fr = af.alignPanel.getFeatureRenderer();
 451     Map<String, Object> visible = fr
 452             .getDisplayedFeatureCols();
 453     String exported = featuresFile.printJalviewFormat(
 454             al.getSequencesArray(), visible);
 455     String expected = "No Features Visible";
 456     assertEquals(expected, exported);
 457
 458     /*
 459      * set METAL (in uniprot group) and GAMMA-TURN visible, but not Pfam
 460      */
 461     fr.setVisible("METAL");
 462     fr.setVisible("GAMMA-TURN");
 463     visible = fr.getDisplayedFeatureCols();
 464     exported = featuresFile.printJalviewFormat(al.getSequencesArray(),
 465             visible);
 466     expected = "METAL\tcc9900\n"
 467             + "GAMMA-TURN\tff0000|00ffff|20.0|95.0|below|66.0\n"
 468             + "\nSTARTGROUP\tuniprot\n"
 469             + "Iron\tFER_CAPAA\t-1\t39\t39\tMETAL\t0.0\n"
 470             + "Turn\tFER_CAPAA\t-1\t36\t38\tGAMMA-TURN\t0.0\n"
 471             + "ENDGROUP\tuniprot\n";
 472     assertEquals(expected, exported);
 473
 474     /*
 475      * now set Pfam visible
 476      */
 477     fr.setVisible("Pfam");
 478     visible = fr.getDisplayedFeatureCols();
 479     exported = featuresFile.printJalviewFormat(al.getSequencesArray(),
 480             visible);
 481     /*
 482      * note the order of feature types is uncontrolled - derives from
 483      * FeaturesDisplayed.featuresDisplayed which is a HashSet
 484      */
 485     expected = "METAL\tcc9900\n"
 486             + "Pfam\tff0000\n"
 487             + "GAMMA-TURN\tff0000|00ffff|20.0|95.0|below|66.0\n"
 488             + "\nSTARTGROUP\tuniprot\n"
 489             + "Iron\tFER_CAPAA\t-1\t39\t39\tMETAL\t0.0\n"
 490             + "Turn\tFER_CAPAA\t-1\t36\t38\tGAMMA-TURN\t0.0\n"
 491             + "<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"
 492             + "ENDGROUP\tuniprot\n";
 493     assertEquals(expected, exported);
 494   }
 495 }