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