test/jalview/io/StockholmFileTest.java

   1 /*
   2  * Jalview - A Sequence Alignment Editor and Viewer (Version 2.9)
   3  * Copyright (C) 2015 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.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.testng.annotations.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(groups = { "Functional" })
  47   public void pfamFileIO() throws Exception
  48   {
  49     testFileIOwithFormat(new File(PfamFile), "STH", -1, 0);
  50   }
  51
  52   @Test(groups = { "Functional" })
  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.getAllPDBEntries() != null)
  62       {
  63         numpdb += sq.getAllPDBEntries().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(groups = { "Functional" })
  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
  88   public static void testFileIOwithFormat(File f, String ioformat,
  89           int naliannot, int nminseqann)
  90   {
  91     System.out.println("Reading file: " + f);
  92     String ff = f.getPath();
  93     try
  94     {
  95       AppletFormatAdapter rf = new AppletFormatAdapter();
  96
  97       AlignmentI al = rf.readFile(ff, AppletFormatAdapter.FILE,
  98               new IdentifyFile().Identify(ff, AppletFormatAdapter.FILE));
  99
 100       assertNotNull("Couldn't read supplied alignment data.", al);
 101
 102       // make sure dataset is initialised ? not sure about this
 103       for (int i = 0; i < al.getSequencesArray().length; ++i)
 104       {
 105         al.getSequenceAt(i).setDatasetSequence(al.getSequenceAt(i));
 106       }
 107       String outputfile = rf.formatSequences(ioformat, al, true);
 108       System.out.println("Output file in '" + ioformat + "':\n"
 109               + outputfile + "\n<<EOF\n");
 110       // test for consistency in io
 111       AlignmentI al_input = new AppletFormatAdapter().readFile(outputfile,
 112               AppletFormatAdapter.PASTE, ioformat);
 113       assertNotNull("Couldn't parse reimported alignment data.", al_input);
 114
 115       String identifyoutput = new IdentifyFile().Identify(outputfile,
 116               AppletFormatAdapter.PASTE);
 117       assertNotNull("Identify routine failed for outputformat " + ioformat,
 118               identifyoutput);
 119       assertTrue(
 120               "Identify routine could not recognise output generated by '"
 121                       + ioformat + "' writer",
 122               ioformat.equals(identifyoutput));
 123       testAlignmentEquivalence(al, al_input, false);
 124       int numaliannot = 0, numsqswithali = 0;
 125       for (AlignmentAnnotation ala : al_input.getAlignmentAnnotation())
 126       {
 127         if (ala.sequenceRef == null)
 128         {
 129           numaliannot++;
 130         }
 131         else
 132         {
 133           numsqswithali++;
 134         }
 135       }
 136       if (naliannot > -1)
 137       {
 138         assertEquals("Number of alignment annotations", naliannot,
 139                 numaliannot);
 140       }
 141
 142       assertTrue(
 143               "Number of sequence associated annotations wasn't at least "
 144                       + nminseqann, numsqswithali >= nminseqann);
 145
 146     } catch (Exception e)
 147     {
 148       e.printStackTrace();
 149       assertTrue("Couln't format the alignment for output file.", false);
 150     }
 151   }
 152
 153   /**
 154    * assert alignment equivalence
 155    *
 156    * @param al
 157    *          'original'
 158    * @param al_input
 159    *          'secondary' or generated alignment from some datapreserving
 160    *          transformation
 161    * @param ignoreFeatures
 162    *          when true, differences in seuqence feature annotation are ignored.
 163    */
 164   public static void testAlignmentEquivalence(AlignmentI al,
 165           AlignmentI al_input, boolean ignoreFeatures)
 166   {
 167     assertNotNull("Original alignment was null", al);
 168     assertNotNull("Generated alignment was null", al_input);
 169
 170     assertTrue(
 171             "Alignment dimension mismatch: originl contains "
 172                     + al.getHeight() + " and generated has "
 173                     + al_input.getHeight() + " sequences; original has "
 174                     + al.getWidth() + " and generated has "
 175                     + al_input.getWidth() + " columns.",
 176             al.getHeight() == al_input.getHeight()
 177                     && al.getWidth() == al_input.getWidth());
 178
 179     // check Alignment annotation
 180     AlignmentAnnotation[] aa_new = al_input.getAlignmentAnnotation();
 181     AlignmentAnnotation[] aa_original = al.getAlignmentAnnotation();
 182
 183     // note - at moment we do not distinguish between alignment without any
 184     // annotation rows and alignment with no annotation row vector
 185     // we might want to revise this in future
 186     int aa_new_size = (aa_new == null ? 0 : aa_new.length), aa_original_size = (aa_original == null ? 0
 187             : aa_original.length);
 188     Map<Integer, java.util.BitSet> orig_groups = new HashMap<Integer, java.util.BitSet>(), new_groups = new HashMap<Integer, java.util.BitSet>();
 189
 190     if (aa_new != null && aa_original != null)
 191     {
 192       for (int i = 0; i < aa_original.length; i++)
 193       {
 194         if (aa_new.length > i)
 195         {
 196           assertTrue("Different alignment annotation at position " + i,
 197                   equalss(aa_original[i], aa_new[i]));
 198           // compare graphGroup or graph properties - needed to verify JAL-1299
 199           assertTrue("Graph type not identical.",
 200                   aa_original[i].graph == aa_new[i].graph);
 201           assertTrue("Visibility not identical.",
 202                   aa_original[i].visible == aa_new[i].visible);
 203           assertTrue(
 204                   "Threshold line not identical.",
 205                   aa_original[i].threshold == null ? aa_new[i].threshold == null
 206                           : aa_original[i].threshold
 207                                   .equals(aa_new[i].threshold));
 208           // graphGroup may differ, but pattern should be the same
 209           Integer o_ggrp = new Integer(aa_original[i].graphGroup + 2), n_ggrp = new Integer(
 210                   aa_new[i].graphGroup + 2);
 211           BitSet orig_g = orig_groups.get(o_ggrp), new_g = new_groups
 212                   .get(n_ggrp);
 213           if (orig_g == null)
 214           {
 215             orig_groups.put(o_ggrp, orig_g = new BitSet());
 216           }
 217           if (new_g == null)
 218           {
 219             new_groups.put(n_ggrp, new_g = new BitSet());
 220           }
 221           assertTrue("Graph Group pattern differs at annotation " + i,
 222                   orig_g.equals(new_g));
 223           orig_g.set(i);
 224           new_g.set(i);
 225         }
 226         else
 227         {
 228           System.err.println("No matching annotation row for "
 229                   + aa_original[i].toString());
 230         }
 231       }
 232     }
 233     assertTrue(
 234             "Generated and imported alignment have different annotation sets ("
 235                     + aa_new_size + " != " + aa_original_size + ")",
 236             aa_new_size == aa_original_size);
 237
 238     // check sequences, annotation and features
 239     SequenceI[] seq_original = new SequenceI[al.getSequencesArray().length];
 240     seq_original = al.getSequencesArray();
 241     SequenceI[] seq_new = new SequenceI[al_input.getSequencesArray().length];
 242     seq_new = al_input.getSequencesArray();
 243     SequenceFeature[] sequenceFeatures_original, sequenceFeatures_new;
 244     AlignmentAnnotation annot_original, annot_new;
 245     //
 246     for (int i = 0; i < al.getSequencesArray().length; i++)
 247     {
 248       String name = seq_original[i].getName();
 249       int start = seq_original[i].getStart();
 250       int end = seq_original[i].getEnd();
 251       System.out.println("Check sequence: " + name + "/" + start + "-"
 252               + end);
 253
 254       // search equal sequence
 255       for (int in = 0; in < al_input.getSequencesArray().length; in++)
 256       {
 257         if (name.equals(seq_new[in].getName())
 258                 && start == seq_new[in].getStart()
 259                 && end == seq_new[in].getEnd())
 260         {
 261           String ss_original = seq_original[i].getSequenceAsString();
 262           String ss_new = seq_new[in].getSequenceAsString();
 263           assertTrue("The sequences " + name + "/" + start + "-" + end
 264                   + " are not equal", ss_original.equals(ss_new));
 265
 266           assertTrue(
 267                   "Sequence Features were not equivalent"
 268                           + (ignoreFeatures ? " ignoring." : ""),
 269                   ignoreFeatures
 270                           || (seq_original[i].getSequenceFeatures() == null && seq_new[in]
 271                                   .getSequenceFeatures() == null)
 272                           || (seq_original[i].getSequenceFeatures() != null && seq_new[in]
 273                                   .getSequenceFeatures() != null));
 274           // compare sequence features
 275           if (seq_original[i].getSequenceFeatures() != null
 276                   && seq_new[in].getSequenceFeatures() != null)
 277           {
 278             System.out.println("There are feature!!!");
 279             sequenceFeatures_original = new SequenceFeature[seq_original[i]
 280                     .getSequenceFeatures().length];
 281             sequenceFeatures_original = seq_original[i]
 282                     .getSequenceFeatures();
 283             sequenceFeatures_new = new SequenceFeature[seq_new[in]
 284                     .getSequenceFeatures().length];
 285             sequenceFeatures_new = seq_new[in].getSequenceFeatures();
 286
 287             assertTrue("different number of features", seq_original[i]
 288                     .getSequenceFeatures().length == seq_new[in]
 289                     .getSequenceFeatures().length);
 290
 291             for (int feat = 0; feat < seq_original[i].getSequenceFeatures().length; feat++)
 292             {
 293               assertTrue("Different features",
 294                       sequenceFeatures_original[feat]
 295                               .equals(sequenceFeatures_new[feat]));
 296             }
 297           }
 298           // compare alignment annotation
 299           if (al.getSequenceAt(i).getAnnotation() != null
 300                   && al_input.getSequenceAt(in).getAnnotation() != null)
 301           {
 302             for (int j = 0; j < al.getSequenceAt(i).getAnnotation().length; j++)
 303             {
 304               if (al.getSequenceAt(i).getAnnotation()[j] != null
 305                       && al_input.getSequenceAt(in).getAnnotation()[j] != null)
 306               {
 307                 annot_original = al.getSequenceAt(i).getAnnotation()[j];
 308                 annot_new = al_input.getSequenceAt(in).getAnnotation()[j];
 309                 assertTrue("Different annotation elements",
 310                         equalss(annot_original, annot_new));
 311               }
 312             }
 313           }
 314           else if (al.getSequenceAt(i).getAnnotation() == null
 315                   && al_input.getSequenceAt(in).getAnnotation() == null)
 316           {
 317             System.out.println("No annotations");
 318           }
 319           else if (al.getSequenceAt(i).getAnnotation() != null
 320                   && al_input.getSequenceAt(in).getAnnotation() == null)
 321           {
 322             assertTrue("Annotations differed between sequences ("
 323                     + al.getSequenceAt(i).getName() + ") and ("
 324                     + al_input.getSequenceAt(i).getName() + ")", false);
 325           }
 326           break;
 327         }
 328       }
 329     }
 330   }
 331
 332   /*
 333    * compare annotations
 334    */
 335   private static boolean equalss(AlignmentAnnotation annot_or,
 336           AlignmentAnnotation annot_new)
 337   {
 338     if (annot_or.annotations.length != annot_new.annotations.length)
 339     {
 340       System.err.println("Different lengths for annotation row elements: "
 341               + annot_or.annotations.length + "!="
 342               + annot_new.annotations.length);
 343       return false;
 344     }
 345     for (int i = 0; i < annot_or.annotations.length; i++)
 346     {
 347       Annotation an_or = annot_or.annotations[i], an_new = annot_new.annotations[i];
 348       if (an_or != null && an_new != null)
 349       {
 350         if (!an_or.displayCharacter.trim().equals(
 351                 an_new.displayCharacter.trim())
 352                 || !("" + an_or.secondaryStructure).trim().equals(
 353                         ("" + an_new.secondaryStructure).trim())
 354                 || (an_or.description != an_new.description && !((an_or.description == null && an_new.description
 355                         .trim().length() == 0)
 356                         || (an_new.description == null && an_or.description
 357                                 .trim().length() == 0) || an_or.description
 358                         .trim().equals(an_new.description.trim()))))
 359         {
 360           System.err.println("Annotation Element Mismatch\nElement " + i
 361                   + " in original: " + annot_or.annotations[i].toString()
 362                   + "\nElement " + i + " in new: "
 363                   + annot_new.annotations[i].toString());
 364           return false;
 365         }
 366       }
 367       else if (annot_or.annotations[i] == null
 368               && annot_new.annotations[i] == null)
 369       {
 370         continue;
 371       }
 372       else
 373       {
 374         System.err.println("Annotation Element Mismatch\nElement "
 375                 + i
 376                 + " in original: "
 377                 + (annot_or.annotations[i] == null ? "is null"
 378                         : annot_or.annotations[i].toString())
 379                 + "\nElement "
 380                 + i
 381                 + " in new: "
 382                 + (annot_new.annotations[i] == null ? "is null"
 383                         : annot_new.annotations[i].toString()));
 384         return false;
 385       }
 386     }
 387     return true;
 388   }
 389 }