test/jalview/io/StockholmFileTest.java

   1 /*
   2  * Jalview - A Sequence Alignment Editor and Viewer (Version 2.8.2b1)
   3  * Copyright (C) 2014 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.junit.Assert.assertEquals;
  24 import static org.junit.Assert.assertNotNull;
  25 import static org.junit.Assert.assertTrue;
  26 import jalview.datamodel.Alignment;
  27 import jalview.datamodel.AlignmentAnnotation;
  28 import jalview.datamodel.AlignmentI;
  29 import jalview.datamodel.Annotation;
  30 import jalview.datamodel.SequenceFeature;
  31 import jalview.datamodel.SequenceI;
  32
  33 import java.io.File;
  34 import java.util.BitSet;
  35 import java.util.HashMap;
  36 import java.util.Map;
  37
  38 import org.junit.Test;
  39
  40 public class StockholmFileTest
  41 {
  42
  43   static String PfamFile = "examples/PF00111_seed.stk",
  44           RfamFile = "examples/RF00031_folded.stk";
  45
  46   @Test
  47   public void pfamFileIO() throws Exception
  48   {
  49     testFileIOwithFormat(new File(PfamFile), "STH", -1, 0);
  50   }
  51
  52   @Test
  53   public void pfamFileDataExtraction() throws Exception
  54   {
  55     AppletFormatAdapter af = new AppletFormatAdapter();
  56     AlignmentI al = af.readFile(PfamFile, af.FILE,
  57             new IdentifyFile().Identify(PfamFile, af.FILE));
  58     int numpdb = 0;
  59     for (SequenceI sq : al.getSequences())
  60     {
  61       if (sq.getPDBId() != null)
  62       {
  63         numpdb += sq.getPDBId().size();
  64       }
  65     }
  66     assertTrue(
  67             "PF00111 seed alignment has at least 1 PDB file, but the reader found none.",
  68             numpdb > 0);
  69   }
  70
  71   @Test
  72   public void rfamFileIO() throws Exception
  73   {
  74     testFileIOwithFormat(new File(RfamFile), "STH", 2, 1);
  75   }
  76
  77   /**
  78    * test alignment data in given file can be imported, exported and reimported
  79    * with no dataloss
  80    *
  81    * @param f
  82    *          - source datafile (IdentifyFile.identify() should work with it)
  83    * @param ioformat
  84    *          - label for IO class used to write and read back in the data from
  85    *          f
  86    */
  87   public static void testFileIOwithFormat(File f, String ioformat,
  88           int naliannot, int nminseqann)
  89   {
  90     System.out.println("Reading file: " + f);
  91     String ff = f.getPath();
  92     try
  93     {
  94       AppletFormatAdapter rf = new AppletFormatAdapter();
  95
  96       Alignment al = rf.readFile(ff, AppletFormatAdapter.FILE,
  97               new IdentifyFile().Identify(ff, AppletFormatAdapter.FILE));
  98
  99       assertNotNull("Couldn't read supplied alignment data.", al);
 100
 101       // make sure dataset is initialised ? not sure about this
 102       for (int i = 0; i < al.getSequencesArray().length; ++i)
 103       {
 104         al.getSequenceAt(i).setDatasetSequence(al.getSequenceAt(i));
 105       }
 106       String outputfile = rf.formatSequences(ioformat, al, true);
 107       System.out.println("Output file in '" + ioformat + "':\n"
 108               + outputfile + "\n<<EOF\n");
 109       // test for consistency in io
 110       Alignment al_input = new AppletFormatAdapter().readFile(outputfile,
 111               AppletFormatAdapter.PASTE, ioformat);
 112       assertNotNull("Couldn't parse reimported alignment data.", al_input);
 113
 114       String identifyoutput = new IdentifyFile().Identify(outputfile,
 115               AppletFormatAdapter.PASTE);
 116       assertNotNull("Identify routine failed for outputformat " + ioformat,
 117               identifyoutput);
 118       assertTrue(
 119               "Identify routine could not recognise output generated by '"
 120                       + ioformat + "' writer",
 121               ioformat.equals(identifyoutput));
 122       testAlignmentEquivalence(al, al_input);
 123       int numaliannot = 0, numsqswithali = 0;
 124       for (AlignmentAnnotation ala : al_input.getAlignmentAnnotation())
 125       {
 126         if (ala.sequenceRef == null)
 127         {
 128           numaliannot++;
 129         }
 130         else
 131         {
 132           numsqswithali++;
 133         }
 134       }
 135       if (naliannot > -1)
 136       {
 137         assertEquals("Number of alignment annotations", naliannot,
 138               numaliannot);
 139       }
 140
 141       assertTrue(
 142               "Number of sequence associated annotations wasn't at least "
 143                       + nminseqann, numsqswithali >= nminseqann);
 144
 145     } catch (Exception e)
 146     {
 147       e.printStackTrace();
 148       assertTrue("Couln't format the alignment for output file.", false);
 149     }
 150   }
 151
 152   /**
 153    * assert alignment equivalence
 154    *
 155    * @param al
 156    *          'original'
 157    * @param al_input
 158    *          'secondary' or generated alignment from some datapreserving
 159    *          transformation
 160    */
 161   public static void testAlignmentEquivalence(AlignmentI al,
 162           AlignmentI al_input)
 163   {
 164     assertNotNull("Original alignment was null", al);
 165     assertNotNull("Generated alignment was null", al_input);
 166
 167     assertTrue(
 168             "Alignment dimension mismatch: originl contains "
 169                     + al.getHeight() + " and generated has "
 170                     + al_input.getHeight() + " sequences; original has "
 171                     + al.getWidth() + " and generated has "
 172                     + al_input.getWidth() + " columns.",
 173             al.getHeight() == al_input.getHeight()
 174                     && al.getWidth() == al_input.getWidth());
 175
 176     // check Alignment annotation
 177     AlignmentAnnotation[] aa_new = al_input.getAlignmentAnnotation();
 178     AlignmentAnnotation[] aa_original = al.getAlignmentAnnotation();
 179
 180     // note - at moment we do not distinguish between alignment without any
 181     // annotation rows and alignment with no annotation row vector
 182     // we might want to revise this in future
 183     int aa_new_size = (aa_new == null ? 0 : aa_new.length), aa_original_size = (aa_original == null ? 0
 184             : aa_original.length);
 185     Map<Integer, java.util.BitSet> orig_groups = new HashMap<Integer, java.util.BitSet>(), new_groups = new HashMap<Integer, java.util.BitSet>();
 186
 187     if (aa_new != null && aa_original != null)
 188     {
 189       for (int i = 0; i < aa_original.length; i++)
 190       {
 191         if (aa_new.length > i)
 192         {
 193           assertTrue("Different alignment annotation at position " + i,
 194                   equalss(aa_original[i], aa_new[i]));
 195           // compare graphGroup or graph properties - needed to verify JAL-1299
 196           assertTrue("Graph type not identical.",
 197                   aa_original[i].graph == aa_new[i].graph);
 198           assertTrue("Visibility not identical.",
 199                   aa_original[i].visible == aa_new[i].visible);
 200           assertTrue(
 201                   "Threshold line not identical.",
 202                   aa_original[i].threshold == null ? aa_new[i].threshold == null
 203                           : aa_original[i].threshold
 204                                   .equals(aa_new[i].threshold));
 205           // graphGroup may differ, but pattern should be the same
 206           Integer o_ggrp = new Integer(aa_original[i].graphGroup + 2), n_ggrp = new Integer(
 207                   aa_new[i].graphGroup + 2);
 208           BitSet orig_g = orig_groups.get(o_ggrp), new_g = new_groups
 209                   .get(n_ggrp);
 210           if (orig_g == null)
 211           {
 212             orig_groups.put(o_ggrp, orig_g = new BitSet());
 213           }
 214           if (new_g == null)
 215           {
 216             new_groups.put(n_ggrp, new_g = new BitSet());
 217           }
 218           assertTrue("Graph Group pattern differs at annotation " + i,
 219                   orig_g.equals(new_g));
 220           orig_g.set(i);
 221           new_g.set(i);
 222         }
 223         else
 224         {
 225           System.err.println("No matching annotation row for "
 226                   + aa_original[i].toString());
 227         }
 228       }
 229     }
 230     assertTrue(
 231             "Generated and imported alignment have different annotation sets ("
 232                     + aa_new_size + " != " + aa_original_size + ")",
 233             aa_new_size == aa_original_size);
 234
 235     // check sequences, annotation and features
 236     SequenceI[] seq_original = new SequenceI[al.getSequencesArray().length];
 237     seq_original = al.getSequencesArray();
 238     SequenceI[] seq_new = new SequenceI[al_input.getSequencesArray().length];
 239     seq_new = al_input.getSequencesArray();
 240     SequenceFeature[] sequenceFeatures_original, sequenceFeatures_new;
 241     AlignmentAnnotation annot_original, annot_new;
 242     //
 243     for (int i = 0; i < al.getSequencesArray().length; i++)
 244     {
 245       String name = seq_original[i].getName();
 246       int start = seq_original[i].getStart();
 247       int end = seq_original[i].getEnd();
 248       System.out.println("Check sequence: " + name + "/" + start + "-"
 249               + end);
 250
 251       // search equal sequence
 252       for (int in = 0; in < al_input.getSequencesArray().length; in++)
 253       {
 254         if (name.equals(seq_new[in].getName())
 255                 && start == seq_new[in].getStart()
 256                 && end == seq_new[in].getEnd())
 257         {
 258           String ss_original = seq_original[i].getSequenceAsString();
 259           String ss_new = seq_new[in].getSequenceAsString();
 260           assertTrue("The sequences " + name + "/" + start + "-" + end
 261                   + " are not equal", ss_original.equals(ss_new));
 262
 263           assertTrue(
 264                   "Sequence Features were not equivalent",
 265                   (seq_original[i].getSequenceFeatures() == null && seq_new[in]
 266                           .getSequenceFeatures() == null)
 267                           || (seq_original[i].getSequenceFeatures() != null && seq_new[in]
 268                                   .getSequenceFeatures() != null));
 269           // compare sequence features
 270           if (seq_original[i].getSequenceFeatures() != null
 271                   && seq_new[in].getSequenceFeatures() != null)
 272           {
 273             System.out.println("There are feature!!!");
 274             sequenceFeatures_original = new SequenceFeature[seq_original[i]
 275                     .getSequenceFeatures().length];
 276             sequenceFeatures_original = seq_original[i]
 277                     .getSequenceFeatures();
 278             sequenceFeatures_new = new SequenceFeature[seq_new[in]
 279                     .getSequenceFeatures().length];
 280             sequenceFeatures_new = seq_new[in].getSequenceFeatures();
 281
 282             assertTrue("different number of features", seq_original[i]
 283                     .getSequenceFeatures().length == seq_new[in]
 284                     .getSequenceFeatures().length);
 285
 286             for (int feat = 0; feat < seq_original[i].getSequenceFeatures().length; feat++)
 287             {
 288               assertTrue("Different features",
 289                       sequenceFeatures_original[feat]
 290                               .equals(sequenceFeatures_new[feat]));
 291             }
 292           }
 293           // compare alignment annotation
 294           if (al.getSequenceAt(i).getAnnotation() != null
 295                   && al_input.getSequenceAt(in).getAnnotation() != null)
 296           {
 297             for (int j = 0; j < al.getSequenceAt(i).getAnnotation().length; j++)
 298             {
 299               if (al.getSequenceAt(i).getAnnotation()[j] != null
 300                       && al_input.getSequenceAt(in).getAnnotation()[j] != null)
 301               {
 302                 annot_original = al.getSequenceAt(i).getAnnotation()[j];
 303                 annot_new = al_input.getSequenceAt(in).getAnnotation()[j];
 304                 assertTrue("Different annotation elements",
 305                         equalss(annot_original, annot_new));
 306               }
 307             }
 308           }
 309           else if (al.getSequenceAt(i).getAnnotation() == null
 310                   && al_input.getSequenceAt(in).getAnnotation() == null)
 311           {
 312             System.out.println("No annotations");
 313           }
 314           else if (al.getSequenceAt(i).getAnnotation() != null
 315                   && al_input.getSequenceAt(in).getAnnotation() == null)
 316           {
 317             assertTrue("Annotations differed between sequences ("
 318                     + al.getSequenceAt(i).getName() + ") and ("
 319                     + al_input.getSequenceAt(i).getName() + ")", false);
 320           }
 321           break;
 322         }
 323       }
 324     }
 325   }
 326
 327   /*
 328    * compare annotations
 329    */
 330   private static boolean equalss(AlignmentAnnotation annot_or,
 331           AlignmentAnnotation annot_new)
 332   {
 333     if (annot_or.annotations.length != annot_new.annotations.length)
 334     {
 335       System.err.println("Different lengths for annotation row elements: "
 336               + annot_or.annotations.length + "!="
 337               + annot_new.annotations.length);
 338       return false;
 339     }
 340     for (int i = 0; i < annot_or.annotations.length; i++)
 341     {
 342       Annotation an_or = annot_or.annotations[i], an_new = annot_new.annotations[i];
 343       if (an_or != null && an_new != null)
 344       {
 345         if (!an_or.displayCharacter.trim().equals(
 346                 an_new.displayCharacter.trim())
 347                 || !("" + an_or.secondaryStructure).trim().equals(
 348                         ("" + an_new.secondaryStructure).trim())
 349                 || (an_or.description != an_new.description && (an_or.description == null
 350                         || an_new.description == null || !an_or.description
 351                           .equals(an_new.description))))
 352         {
 353           System.err.println("Annotation Element Mismatch\nElement " + i
 354                   + " in original: " + annot_or.annotations[i].toString()
 355                   + "\nElement " + i + " in new: "
 356                   + annot_new.annotations[i].toString());
 357           return false;
 358         }
 359       }
 360       else if (annot_or.annotations[i] == null
 361               && annot_new.annotations[i] == null)
 362       {
 363         continue;
 364       }
 365       else
 366       {
 367         System.err.println("Annotation Element Mismatch\nElement "
 368                 + i
 369                 + " in original: "
 370                 + (annot_or.annotations[i] == null ? "is null"
 371                         : annot_or.annotations[i].toString())
 372                 + "\nElement "
 373                 + i
 374                 + " in new: "
 375                 + (annot_new.annotations[i] == null ? "is null"
 376                         : annot_new.annotations[i].toString()));
 377         return false;
 378       }
 379     }
 380     return true;
 381   }
 382 }