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