test/jalview/io/StockholmFileTest.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.assertNotNull;
  25 import static org.testng.AssertJUnit.assertTrue;
  26 import static org.testng.AssertJUnit.fail;
  27
  28 import jalview.datamodel.AlignmentAnnotation;
  29 import jalview.datamodel.AlignmentI;
  30 import jalview.datamodel.Annotation;
  31 import jalview.datamodel.SequenceFeature;
  32 import jalview.datamodel.SequenceI;
  33 import jalview.gui.JvOptionPane;
  34
  35 import java.io.File;
  36 import java.util.BitSet;
  37 import java.util.HashMap;
  38 import java.util.Map;
  39
  40 import org.testng.annotations.BeforeClass;
  41 import org.testng.annotations.Test;
  42
  43 public class StockholmFileTest
  44 {
  45
  46   @BeforeClass(alwaysRun = true)
  47   public void setUpJvOptionPane()
  48   {
  49     JvOptionPane.setInteractiveMode(false);
  50     JvOptionPane.setMockResponse(JvOptionPane.CANCEL_OPTION);
  51   }
  52
  53   static String PfamFile = "examples/PF00111_seed.stk",
  54           RfamFile = "examples/RF00031_folded.stk";
  55
  56   @Test(groups = { "Functional" })
  57   public void pfamFileIO() throws Exception
  58   {
  59     testFileIOwithFormat(new File(PfamFile), FileFormat.Stockholm, -1, 0,
  60             false);
  61   }
  62
  63   @Test(groups = { "Functional" })
  64   public void pfamFileDataExtraction() throws Exception
  65   {
  66     AppletFormatAdapter af = new AppletFormatAdapter();
  67     AlignmentI al = af.readFile(PfamFile, DataSourceType.FILE,
  68             new IdentifyFile().identify(PfamFile, DataSourceType.FILE));
  69     int numpdb = 0;
  70     for (SequenceI sq : al.getSequences())
  71     {
  72       if (sq.getAllPDBEntries() != null)
  73       {
  74         numpdb += sq.getAllPDBEntries().size();
  75       }
  76     }
  77     assertTrue(
  78             "PF00111 seed alignment has at least 1 PDB file, but the reader found none.",
  79             numpdb > 0);
  80   }
  81
  82   @Test(groups = { "Functional" })
  83   public void rfamFileIO() throws Exception
  84   {
  85     testFileIOwithFormat(new File(RfamFile), FileFormat.Stockholm, 2, 1,
  86             false);
  87   }
  88
  89   /**
  90    * test alignment data in given file can be imported, exported and reimported
  91    * with no dataloss
  92    *
  93    * @param f
  94    *          - source datafile (IdentifyFile.identify() should work with it)
  95    * @param ioformat
  96    *          - label for IO class used to write and read back in the data from
  97    *          f
  98    * @param ignoreRowVisibility
  99    */
 100
 101   public static void testFileIOwithFormat(File f, FileFormatI ioformat,
 102           int naliannot, int nminseqann, boolean ignoreRowVisibility)
 103   {
 104     System.out.println("Reading file: " + f);
 105     String ff = f.getPath();
 106     try
 107     {
 108       AppletFormatAdapter rf = new AppletFormatAdapter();
 109
 110       AlignmentI al = rf.readFile(ff, DataSourceType.FILE,
 111               new IdentifyFile().identify(ff, DataSourceType.FILE));
 112
 113       assertNotNull("Couldn't read supplied alignment data.", al);
 114
 115       // make sure dataset is initialised ? not sure about this
 116       for (int i = 0; i < al.getSequencesArray().length; ++i)
 117       {
 118         al.getSequenceAt(i).createDatasetSequence();
 119       }
 120       String outputfile = rf.formatSequences(ioformat, al, true);
 121       System.out.println("Output file in '" + ioformat + "':\n"
 122               + outputfile + "\n<<EOF\n");
 123       // test for consistency in io
 124       AlignmentI al_input = new AppletFormatAdapter().readFile(outputfile,
 125               DataSourceType.PASTE, ioformat);
 126       assertNotNull("Couldn't parse reimported alignment data.", al_input);
 127
 128       FileFormatI identifyoutput = new IdentifyFile().identify(outputfile,
 129               DataSourceType.PASTE);
 130       assertNotNull("Identify routine failed for outputformat " + ioformat,
 131               identifyoutput);
 132       assertTrue(
 133               "Identify routine could not recognise output generated by '"
 134                       + ioformat + "' writer",
 135               ioformat.equals(identifyoutput));
 136       testAlignmentEquivalence(al, al_input, false);
 137       int numaliannot = 0, numsqswithali = 0;
 138       for (AlignmentAnnotation ala : al_input.getAlignmentAnnotation())
 139       {
 140         if (ala.sequenceRef == null)
 141         {
 142           numaliannot++;
 143         }
 144         else
 145         {
 146           numsqswithali++;
 147         }
 148       }
 149       if (naliannot > -1)
 150       {
 151         assertEquals("Number of alignment annotations", naliannot,
 152                 numaliannot);
 153       }
 154
 155       assertTrue(
 156               "Number of sequence associated annotations wasn't at least "
 157                       + nminseqann, numsqswithali >= nminseqann);
 158
 159     } catch (Exception e)
 160     {
 161       e.printStackTrace();
 162       assertTrue("Couln't format the alignment for output file.", false);
 163     }
 164   }
 165
 166   /**
 167    * assert alignment equivalence
 168    *
 169    * @param al
 170    *          'original'
 171    * @param al_input
 172    *          'secondary' or generated alignment from some datapreserving
 173    *          transformation
 174    * @param ignoreFeatures
 175    *          when true, differences in sequence feature annotation are ignored
 176    */
 177   public static void testAlignmentEquivalence(AlignmentI al,
 178           AlignmentI al_input, boolean ignoreFeatures)
 179   {
 180   }
 181
 182   /**
 183    * assert alignment equivalence
 184    *
 185    * @param al
 186    *          'original'
 187    * @param al_input
 188    *          'secondary' or generated alignment from some datapreserving
 189    *          transformation
 190    * @param ignoreFeatures
 191    *          when true, differences in sequence feature annotation are ignored
 192    *
 193    * @param ignoreRowVisibility
 194    *          when true, do not fail if there are differences in the visibility
 195    *          of annotation rows
 196    */
 197   public static void testAlignmentEquivalence(AlignmentI al,
 198           AlignmentI al_input, boolean ignoreFeatures,
 199           boolean ignoreRowVisibility)
 200   {
 201     assertNotNull("Original alignment was null", al);
 202     assertNotNull("Generated alignment was null", al_input);
 203
 204     assertTrue("Alignment dimension mismatch: original: " + al.getHeight()
 205             + "x" + al.getWidth() + ", generated: " + al_input.getHeight()
 206             + "x" + al_input.getWidth(),
 207             al.getHeight() == al_input.getHeight()
 208                     && al.getWidth() == al_input.getWidth());
 209
 210     // check Alignment annotation
 211     AlignmentAnnotation[] aa_new = al_input.getAlignmentAnnotation();
 212     AlignmentAnnotation[] aa_original = al.getAlignmentAnnotation();
 213
 214     // note - at moment we do not distinguish between alignment without any
 215     // annotation rows and alignment with no annotation row vector
 216     // we might want to revise this in future
 217     int aa_new_size = (aa_new == null ? 0 : aa_new.length);
 218     int aa_original_size = (aa_original == null ? 0 : aa_original.length);
 219     Map<Integer, BitSet> orig_groups = new HashMap<Integer, BitSet>();
 220     Map<Integer, BitSet> new_groups = new HashMap<Integer, BitSet>();
 221
 222     if (aa_new != null && aa_original != null)
 223     {
 224       for (int i = 0; i < aa_original.length; i++)
 225       {
 226         if (aa_new.length > i)
 227         {
 228           assertEqualSecondaryStructure(
 229                   "Different alignment annotation at position " + i,
 230                   aa_original[i], aa_new[i]);
 231           // compare graphGroup or graph properties - needed to verify JAL-1299
 232           assertEquals("Graph type not identical.", aa_original[i].graph,
 233                   aa_new[i].graph);
 234           if (!ignoreRowVisibility)
 235           {
 236             assertEquals("Visibility not identical.",
 237                     aa_original[i].visible,
 238                   aa_new[i].visible);
 239           }
 240           assertEquals("Threshold line not identical.",
 241                   aa_original[i].threshold, aa_new[i].threshold);
 242           // graphGroup may differ, but pattern should be the same
 243           Integer o_ggrp = new Integer(aa_original[i].graphGroup + 2);
 244           Integer n_ggrp = new Integer(aa_new[i].graphGroup + 2);
 245           BitSet orig_g = orig_groups.get(o_ggrp);
 246           BitSet new_g = new_groups.get(n_ggrp);
 247           if (orig_g == null)
 248           {
 249             orig_groups.put(o_ggrp, orig_g = new BitSet());
 250           }
 251           if (new_g == null)
 252           {
 253             new_groups.put(n_ggrp, new_g = new BitSet());
 254           }
 255           assertEquals("Graph Group pattern differs at annotation " + i,
 256                   orig_g, new_g);
 257           orig_g.set(i);
 258           new_g.set(i);
 259         }
 260         else
 261         {
 262           System.err.println("No matching annotation row for "
 263                   + aa_original[i].toString());
 264         }
 265       }
 266     }
 267     assertEquals(
 268             "Generated and imported alignment have different annotation sets",
 269             aa_original_size, aa_new_size);
 270
 271     // check sequences, annotation and features
 272     SequenceI[] seq_original = new SequenceI[al.getSequencesArray().length];
 273     seq_original = al.getSequencesArray();
 274     SequenceI[] seq_new = new SequenceI[al_input.getSequencesArray().length];
 275     seq_new = al_input.getSequencesArray();
 276     SequenceFeature[] sequenceFeatures_original, sequenceFeatures_new;
 277     AlignmentAnnotation annot_original, annot_new;
 278     //
 279     for (int i = 0; i < al.getSequencesArray().length; i++)
 280     {
 281       String name = seq_original[i].getName();
 282       int start = seq_original[i].getStart();
 283       int end = seq_original[i].getEnd();
 284       System.out.println("Check sequence: " + name + "/" + start + "-"
 285               + end);
 286
 287       // search equal sequence
 288       for (int in = 0; in < al_input.getSequencesArray().length; in++)
 289       {
 290         if (name.equals(seq_new[in].getName())
 291                 && start == seq_new[in].getStart()
 292                 && end == seq_new[in].getEnd())
 293         {
 294           String ss_original = seq_original[i].getSequenceAsString();
 295           String ss_new = seq_new[in].getSequenceAsString();
 296           assertEquals("The sequences " + name + "/" + start + "-" + end
 297                   + " are not equal", ss_original, ss_new);
 298
 299           assertTrue(
 300                   "Sequence Features were not equivalent"
 301                           + (ignoreFeatures ? " ignoring." : ""),
 302                   ignoreFeatures
 303                           || (seq_original[i].getSequenceFeatures() == null && seq_new[in]
 304                                   .getSequenceFeatures() == null)
 305                           || (seq_original[i].getSequenceFeatures() != null && seq_new[in]
 306                                   .getSequenceFeatures() != null));
 307           // compare sequence features
 308           if (seq_original[i].getSequenceFeatures() != null
 309                   && seq_new[in].getSequenceFeatures() != null)
 310           {
 311             System.out.println("There are feature!!!");
 312             sequenceFeatures_original = new SequenceFeature[seq_original[i]
 313                     .getSequenceFeatures().length];
 314             sequenceFeatures_original = seq_original[i]
 315                     .getSequenceFeatures();
 316             sequenceFeatures_new = new SequenceFeature[seq_new[in]
 317                     .getSequenceFeatures().length];
 318             sequenceFeatures_new = seq_new[in].getSequenceFeatures();
 319
 320             assertEquals("different number of features",
 321                     seq_original[i].getSequenceFeatures().length,
 322                     seq_new[in].getSequenceFeatures().length);
 323
 324             for (int feat = 0; feat < seq_original[i].getSequenceFeatures().length; feat++)
 325             {
 326               assertEquals("Different features",
 327                       sequenceFeatures_original[feat],
 328                       sequenceFeatures_new[feat]);
 329             }
 330           }
 331           // compare alignment annotation
 332           if (al.getSequenceAt(i).getAnnotation() != null
 333                   && al_input.getSequenceAt(in).getAnnotation() != null)
 334           {
 335             for (int j = 0; j < al.getSequenceAt(i).getAnnotation().length; j++)
 336             {
 337               if (al.getSequenceAt(i).getAnnotation()[j] != null
 338                       && al_input.getSequenceAt(in).getAnnotation()[j] != null)
 339               {
 340                 annot_original = al.getSequenceAt(i).getAnnotation()[j];
 341                 annot_new = al_input.getSequenceAt(in).getAnnotation()[j];
 342                 assertEqualSecondaryStructure(
 343                         "Different annotation elements", annot_original,
 344                         annot_new);
 345               }
 346             }
 347           }
 348           else if (al.getSequenceAt(i).getAnnotation() == null
 349                   && al_input.getSequenceAt(in).getAnnotation() == null)
 350           {
 351             System.out.println("No annotations");
 352           }
 353           else if (al.getSequenceAt(i).getAnnotation() != null
 354                   && al_input.getSequenceAt(in).getAnnotation() == null)
 355           {
 356             fail("Annotations differed between sequences ("
 357                     + al.getSequenceAt(i).getName() + ") and ("
 358                     + al_input.getSequenceAt(i).getName() + ")");
 359           }
 360           break;
 361         }
 362       }
 363     }
 364   }
 365
 366   private static void assertEqualSecondaryStructure(String message,
 367           AlignmentAnnotation annot_or,
 368           AlignmentAnnotation annot_new)
 369   {
 370     // TODO: test to cover this assert behaves correctly for all allowed
 371     // variations of secondary structure annotation row equivalence
 372     if (annot_or.annotations.length != annot_new.annotations.length)
 373     {
 374       fail("Different lengths for annotation row elements: "
 375               + annot_or.annotations.length + "!="
 376               + annot_new.annotations.length);
 377     }
 378     boolean isRna = annot_or.isRNA();
 379     assertTrue("Expected " + (isRna ? " valid RNA " : " no RNA ")
 380             + " secondary structure in the row.",
 381             isRna == annot_new.isRNA());
 382     for (int i = 0; i < annot_or.annotations.length; i++)
 383     {
 384       Annotation an_or = annot_or.annotations[i], an_new = annot_new.annotations[i];
 385       if (an_or != null && an_new != null)
 386       {
 387
 388         if (isRna)
 389         {
 390           if (an_or.secondaryStructure != an_new.secondaryStructure
 391                   || an_or.value != an_new.value)
 392           {
 393             fail("Different RNA secondary structure at column " + i
 394                     + " expected: [" + annot_or.annotations[i].toString()
 395                     + "] but got: [" + annot_new.annotations[i].toString()
 396                     + "]");
 397           }
 398         }
 399         else
 400         {
 401           // not RNA secondary structure, so expect all elements to match...
 402           if (!an_or.displayCharacter.trim().equals(
 403                   an_new.displayCharacter.trim())
 404                   || !("" + an_or.secondaryStructure).trim().equals(
 405                           ("" + an_new.secondaryStructure).trim())
 406                   || (an_or.description != an_new.description && !((an_or.description == null && an_new.description
 407                           .trim().length() == 0)
 408                           || (an_new.description == null && an_or.description
 409                                   .trim().length() == 0) || an_or.description
 410                           .trim().equals(an_new.description.trim()))))
 411           {
 412             fail("Annotation Element Mismatch\nElement " + i
 413                     + " in original: " + annot_or.annotations[i].toString()
 414                     + "\nElement " + i + " in new: "
 415                     + annot_new.annotations[i].toString());
 416           }
 417         }
 418       }
 419       else if (annot_or.annotations[i] == null
 420               && annot_new.annotations[i] == null)
 421       {
 422         continue;
 423       }
 424       else
 425       {
 426         fail("Annotation Element Mismatch\nElement "
 427                 + i
 428                 + " in original: "
 429                 + (annot_or.annotations[i] == null ? "is null"
 430                         : annot_or.annotations[i].toString())
 431                 + "\nElement "
 432                 + i
 433                 + " in new: "
 434                 + (annot_new.annotations[i] == null ? "is null"
 435                         : annot_new.annotations[i].toString()));
 436       }
 437     }
 438   }
 439 }