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