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
  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(groups = { "Functional" })
  48   public void pfamFileIO() throws Exception
  49   {
  50     testFileIOwithFormat(new File(PfamFile), "STH", -1, 0);
  51   }
  52
  53   @Test(groups = { "Functional" })
  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.getAllPDBEntries() != null)
  63       {
  64         numpdb += sq.getAllPDBEntries().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(groups = { "Functional" })
  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       AlignmentI 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).createDatasetSequence();
 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       AlignmentI 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 sequence 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("Alignment dimension mismatch: original: " + al.getHeight()
 172             + "x" + al.getWidth() + ", generated: " + al_input.getHeight()
 173             + "x" + al_input.getWidth(),
 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);
 185     int aa_original_size = (aa_original == null ? 0 : aa_original.length);
 186     Map<Integer, BitSet> orig_groups = new HashMap<Integer, BitSet>();
 187     Map<Integer, BitSet> new_groups = new HashMap<Integer, BitSet>();
 188
 189     if (aa_new != null && aa_original != null)
 190     {
 191       for (int i = 0; i < aa_original.length; i++)
 192       {
 193         if (aa_new.length > i)
 194         {
 195           assertTrue("Different alignment annotation at position " + i,
 196                   equalss(aa_original[i], aa_new[i]));
 197           // compare graphGroup or graph properties - needed to verify JAL-1299
 198           assertEquals("Graph type not identical.", aa_original[i].graph,
 199                   aa_new[i].graph);
 200           assertEquals("Visibility not identical.", aa_original[i].visible,
 201                   aa_new[i].visible);
 202           assertEquals("Threshold line not identical.",
 203                   aa_original[i].threshold, aa_new[i].threshold);
 204           // graphGroup may differ, but pattern should be the same
 205           Integer o_ggrp = new Integer(aa_original[i].graphGroup + 2);
 206           Integer n_ggrp = new Integer(aa_new[i].graphGroup + 2);
 207           BitSet orig_g = orig_groups.get(o_ggrp);
 208           BitSet new_g = new_groups.get(n_ggrp);
 209           if (orig_g == null)
 210           {
 211             orig_groups.put(o_ggrp, orig_g = new BitSet());
 212           }
 213           if (new_g == null)
 214           {
 215             new_groups.put(n_ggrp, new_g = new BitSet());
 216           }
 217           assertEquals("Graph Group pattern differs at annotation " + i,
 218                   orig_g, new_g);
 219           orig_g.set(i);
 220           new_g.set(i);
 221         }
 222         else
 223         {
 224           System.err.println("No matching annotation row for "
 225                   + aa_original[i].toString());
 226         }
 227       }
 228     }
 229     assertEquals(
 230             "Generated and imported alignment have different annotation sets",
 231             aa_original_size, aa_new_size);
 232
 233     // check sequences, annotation and features
 234     SequenceI[] seq_original = new SequenceI[al.getSequencesArray().length];
 235     seq_original = al.getSequencesArray();
 236     SequenceI[] seq_new = new SequenceI[al_input.getSequencesArray().length];
 237     seq_new = al_input.getSequencesArray();
 238     SequenceFeature[] sequenceFeatures_original, sequenceFeatures_new;
 239     AlignmentAnnotation annot_original, annot_new;
 240     //
 241     for (int i = 0; i < al.getSequencesArray().length; i++)
 242     {
 243       String name = seq_original[i].getName();
 244       int start = seq_original[i].getStart();
 245       int end = seq_original[i].getEnd();
 246       System.out.println("Check sequence: " + name + "/" + start + "-"
 247               + end);
 248
 249       // search equal sequence
 250       for (int in = 0; in < al_input.getSequencesArray().length; in++)
 251       {
 252         if (name.equals(seq_new[in].getName())
 253                 && start == seq_new[in].getStart()
 254                 && end == seq_new[in].getEnd())
 255         {
 256           String ss_original = seq_original[i].getSequenceAsString();
 257           String ss_new = seq_new[in].getSequenceAsString();
 258           assertEquals("The sequences " + name + "/" + start + "-" + end
 259                   + " are not equal", ss_original, ss_new);
 260
 261           assertTrue(
 262                   "Sequence Features were not equivalent"
 263                           + (ignoreFeatures ? " ignoring." : ""),
 264                   ignoreFeatures
 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             assertEquals("different number of features",
 283                     seq_original[i].getSequenceFeatures().length,
 284                     seq_new[in].getSequenceFeatures().length);
 285
 286             for (int feat = 0; feat < seq_original[i].getSequenceFeatures().length; feat++)
 287             {
 288               assertEquals("Different features",
 289                       sequenceFeatures_original[feat],
 290                       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             fail("Annotations differed between sequences ("
 318                     + al.getSequenceAt(i).getName() + ") and ("
 319                     + al_input.getSequenceAt(i).getName() + ")");
 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 && an_new.description
 350                         .trim().length() == 0)
 351                         || (an_new.description == null && an_or.description
 352                                 .trim().length() == 0) || an_or.description
 353                         .trim().equals(an_new.description.trim()))))
 354         {
 355           System.err.println("Annotation Element Mismatch\nElement " + i
 356                   + " in original: " + annot_or.annotations[i].toString()
 357                   + "\nElement " + i + " in new: "
 358                   + annot_new.annotations[i].toString());
 359           return false;
 360         }
 361       }
 362       else if (annot_or.annotations[i] == null
 363               && annot_new.annotations[i] == null)
 364       {
 365         continue;
 366       }
 367       else
 368       {
 369         System.err.println("Annotation Element Mismatch\nElement "
 370                 + i
 371                 + " in original: "
 372                 + (annot_or.annotations[i] == null ? "is null"
 373                         : annot_or.annotations[i].toString())
 374                 + "\nElement "
 375                 + i
 376                 + " in new: "
 377                 + (annot_new.annotations[i] == null ? "is null"
 378                         : annot_new.annotations[i].toString()));
 379         return false;
 380       }
 381     }
 382     return true;
 383   }
 384 }