test/jalview/io/StockholmFileTest.java

   1 /*
   2  * Jalview - A Sequence Alignment Editor and Viewer (Version 2.8.0b1)
   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 of the License, or (at your option) any later version.
  10  *
  11  * Jalview is distributed in the hope that it will be useful, but
  12  * WITHOUT ANY WARRANTY; without even the implied warranty
  13  * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
  14  * PURPOSE.  See the GNU General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU General Public License along with Jalview.  If not, see <http://www.gnu.org/licenses/>.
  17  * The Jalview Authors are detailed in the 'AUTHORS' file.
  18  */
  19 package jalview.io;
  20
  21 import static org.junit.Assert.assertNotNull;
  22 import static org.junit.Assert.assertTrue;
  23 import jalview.datamodel.Alignment;
  24 import jalview.datamodel.AlignmentAnnotation;
  25 import jalview.datamodel.AlignmentI;
  26 import jalview.datamodel.Annotation;
  27 import jalview.datamodel.SequenceFeature;
  28 import jalview.datamodel.SequenceI;
  29
  30 import java.io.File;
  31 import java.util.BitSet;
  32 import java.util.HashMap;
  33 import java.util.Map;
  34
  35 import org.junit.Test;
  36
  37 public class StockholmFileTest
  38 {
  39
  40   static String PfamFile = "examples/PF00111_seed.stk",
  41           RfamFile = "examples/RF00031_folded.stk";
  42
  43   @Test
  44   public void pfamFileIO() throws Exception
  45   {
  46     testFileIOwithFormat(new File(PfamFile), "STH");
  47   }
  48   @Test
  49   public void pfamFileDataExtraction() throws Exception
  50   {  AppletFormatAdapter af = new AppletFormatAdapter();
  51     AlignmentI al = af.readFile(PfamFile, af.FILE,
  52             new IdentifyFile().Identify(PfamFile, af.FILE));
  53     int numpdb = 0;
  54     for (SequenceI sq : al.getSequences())
  55     {
  56       if (sq.getPDBId() != null)
  57       {
  58         numpdb += sq.getPDBId().size();
  59       }
  60     }
  61     assertTrue(
  62             "PF00111 seed alignment has at least 1 PDB file, but the reader found none.",
  63             numpdb > 0);
  64   }
  65
  66   @Test
  67   public void rfamFileIO() throws Exception
  68   {
  69     testFileIOwithFormat(new File(RfamFile), "STH");
  70   }
  71
  72   /**
  73    * test alignment data in given file can be imported, exported and reimported
  74    * with no dataloss
  75    *
  76    * @param f
  77    *          - source datafile (IdentifyFile.identify() should work with it)
  78    * @param ioformat
  79    *          - label for IO class used to write and read back in the data from
  80    *          f
  81    */
  82   public static void testFileIOwithFormat(File f, String ioformat)
  83   {
  84     System.out.println("Reading file: " + f);
  85     String ff = f.getPath();
  86     try
  87     {
  88       AppletFormatAdapter rf = new AppletFormatAdapter();
  89
  90       Alignment al = rf.readFile(ff, AppletFormatAdapter.FILE,
  91               new IdentifyFile().Identify(ff, AppletFormatAdapter.FILE));
  92
  93       assertNotNull("Couldn't read supplied alignment data.", al);
  94
  95       // make sure dataset is initialised ? not sure about this
  96       for (int i = 0; i < al.getSequencesArray().length; ++i)
  97       {
  98         al.getSequenceAt(i).setDatasetSequence(al.getSequenceAt(i));
  99       }
 100       String outputfile = rf.formatSequences(ioformat, al, true);
 101       System.out.println("Output file in '"+ioformat+"':\n"+outputfile+"\n<<EOF\n");
 102       // test for consistency in io
 103       Alignment al_input = new AppletFormatAdapter().readFile(outputfile,
 104               AppletFormatAdapter.PASTE, ioformat);
 105       assertNotNull("Couldn't parse reimported alignment data.", al_input);
 106
 107       String identifyoutput = new IdentifyFile().Identify(outputfile,
 108               AppletFormatAdapter.PASTE);
 109       assertNotNull("Identify routine failed for outputformat " + ioformat,
 110               identifyoutput);
 111       assertTrue(
 112               "Identify routine could not recognise output generated by '"
 113                       + ioformat + "' writer",
 114               ioformat.equals(identifyoutput));
 115       testAlignmentEquivalence(al, al_input);
 116     } catch (Exception e)
 117     {
 118       e.printStackTrace();
 119       assertTrue("Couln't format the alignment for output file.", false);
 120     }
 121   }
 122
 123   /**
 124    * assert alignment equivalence
 125    *
 126    * @param al
 127    *          'original'
 128    * @param al_input
 129    *          'secondary' or generated alignment from some datapreserving
 130    *          transformation
 131    */
 132   public static void testAlignmentEquivalence(AlignmentI al,
 133           AlignmentI al_input)
 134   {
 135     assertNotNull("Original alignment was null", al);
 136     assertNotNull("Generated alignment was null", al_input);
 137
 138     assertTrue(
 139             "Alignment dimension mismatch: originl contains "
 140                     + al.getHeight() + " and generated has "
 141                     + al_input.getHeight() + " sequences; original has "
 142                     + al.getWidth() + " and generated has "
 143                     + al_input.getWidth() + " columns.",
 144             al.getHeight() == al_input.getHeight()
 145                     && al.getWidth() == al_input.getWidth());
 146
 147     // check Alignment annotation
 148     AlignmentAnnotation[] aa_new = al_input.getAlignmentAnnotation();
 149     AlignmentAnnotation[] aa_original = al.getAlignmentAnnotation();
 150
 151     // note - at moment we do not distinguish between alignment without any
 152     // annotation rows and alignment with no annotation row vector
 153     // we might want to revise this in future
 154     int aa_new_size = (aa_new == null ? 0 : aa_new.length), aa_original_size = (aa_original == null ? 0
 155             : aa_original.length);
 156     Map<Integer,java.util.BitSet> orig_groups=new HashMap<Integer,java.util.BitSet>(),new_groups=new HashMap<Integer,java.util.BitSet>();
 157
 158     if (aa_new != null && aa_original != null)
 159     {
 160       for (int i = 0; i < aa_original.length; i++)
 161       {
 162         if (aa_new.length>i) {
 163           assertTrue("Different alignment annotation at position "+i,
 164                 equalss(aa_original[i], aa_new[i]));
 165           // compare graphGroup or graph properties - needed to verify JAL-1299
 166           assertTrue("Graph type not identical.",aa_original[i].graph==aa_new[i].graph);
 167           assertTrue("Visibility not identical.", aa_original[i].visible==aa_new[i].visible);
 168           assertTrue(
 169                   "Threshold line not identical.",
 170                   aa_original[i].threshold == null ? aa_new[i].threshold == null
 171                           : aa_original[i].threshold
 172                                   .equals(aa_new[i].threshold));
 173           // graphGroup may differ, but pattern should be the same
 174           Integer o_ggrp=new Integer(aa_original[i].graphGroup+2),n_ggrp=new Integer(aa_new[i].graphGroup+2);
 175           BitSet orig_g=orig_groups.get(o_ggrp),new_g=new_groups.get(n_ggrp);
 176           if (orig_g==null) {
 177             orig_groups.put(o_ggrp,orig_g= new BitSet());
 178           }
 179           if (new_g==null) {
 180             new_groups.put(n_ggrp, new_g=new BitSet());
 181           }
 182           assertTrue("Graph Group pattern differs at annotation "+i, orig_g.equals(new_g));
 183           orig_g.set(i); new_g.set(i);
 184         } else {
 185           System.err.println("No matching annotation row for "+aa_original[i].toString());
 186         }
 187       }
 188     }
 189     assertTrue(
 190             "Generated and imported alignment have different annotation sets ("
 191                     + aa_new_size + " != " + aa_original_size + ")",
 192             aa_new_size == aa_original_size);
 193
 194     // check sequences, annotation and features
 195     SequenceI[] seq_original = new SequenceI[al.getSequencesArray().length];
 196     seq_original = al.getSequencesArray();
 197     SequenceI[] seq_new = new SequenceI[al_input.getSequencesArray().length];
 198     seq_new = al_input.getSequencesArray();
 199     SequenceFeature[] sequenceFeatures_original, sequenceFeatures_new;
 200     AlignmentAnnotation annot_original, annot_new;
 201     //
 202     for (int i = 0; i < al.getSequencesArray().length; i++)
 203     {
 204       String name = seq_original[i].getName();
 205       int start = seq_original[i].getStart();
 206       int end = seq_original[i].getEnd();
 207       System.out.println("Check sequence: " + name + "/" + start + "-"
 208               + end);
 209
 210       // search equal sequence
 211       for (int in = 0; in < al_input.getSequencesArray().length; in++)
 212       {
 213         if (name.equals(seq_new[in].getName())
 214                 && start == seq_new[in].getStart()
 215                 && end == seq_new[in].getEnd())
 216         {
 217           String ss_original = seq_original[i].getSequenceAsString();
 218           String ss_new = seq_new[in].getSequenceAsString();
 219           assertTrue("The sequences " + name + "/" + start + "-" + end
 220                   + " are not equal", ss_original.equals(ss_new));
 221
 222           assertTrue(
 223                   "Sequence Features were not equivalent",
 224                   (seq_original[i].getSequenceFeatures() == null && seq_new[in]
 225                           .getSequenceFeatures() == null)
 226                           || (seq_original[i].getSequenceFeatures() != null && seq_new[in]
 227                                   .getSequenceFeatures() != null));
 228           // compare sequence features
 229           if (seq_original[i].getSequenceFeatures() != null
 230                   && seq_new[in].getSequenceFeatures() != null)
 231           {
 232             System.out.println("There are feature!!!");
 233             sequenceFeatures_original = new SequenceFeature[seq_original[i]
 234                     .getSequenceFeatures().length];
 235             sequenceFeatures_original = seq_original[i]
 236                     .getSequenceFeatures();
 237             sequenceFeatures_new = new SequenceFeature[seq_new[in]
 238                     .getSequenceFeatures().length];
 239             sequenceFeatures_new = seq_new[in].getSequenceFeatures();
 240
 241             assertTrue("different number of features", seq_original[i]
 242                     .getSequenceFeatures().length == seq_new[in]
 243                     .getSequenceFeatures().length);
 244
 245             for (int feat = 0; feat < seq_original[i].getSequenceFeatures().length; feat++)
 246             {
 247               assertTrue("Different features",
 248                       sequenceFeatures_original[feat]
 249                               .equals(sequenceFeatures_new[feat]));
 250             }
 251           }
 252           // compare alignment annotation
 253           if (al.getSequenceAt(i).getAnnotation() != null
 254                   && al_input.getSequenceAt(in).getAnnotation() != null)
 255           {
 256             for (int j = 0; j < al.getSequenceAt(i).getAnnotation().length; j++)
 257             {
 258               if (al.getSequenceAt(i).getAnnotation()[j] != null
 259                       && al_input.getSequenceAt(in).getAnnotation()[j] != null)
 260               {
 261                 annot_original = al.getSequenceAt(i).getAnnotation()[j];
 262                 annot_new = al_input.getSequenceAt(in).getAnnotation()[j];
 263                 assertTrue("Different annotation elements",
 264                         equalss(annot_original, annot_new));
 265               }
 266             }
 267           }
 268           else if (al.getSequenceAt(i).getAnnotation() == null
 269                   && al_input.getSequenceAt(in).getAnnotation() == null)
 270           {
 271             System.out.println("No annotations");
 272           }
 273           else if (al.getSequenceAt(i).getAnnotation() != null
 274                   && al_input.getSequenceAt(in).getAnnotation() == null)
 275           {
 276             assertTrue("Annotations differed between sequences ("
 277                     + al.getSequenceAt(i).getName() + ") and ("
 278                     + al_input.getSequenceAt(i).getName() + ")", false);
 279           }
 280           break;
 281         }
 282       }
 283     }
 284   }
 285
 286   /*
 287    * compare annotations
 288    */
 289   private static boolean equalss(AlignmentAnnotation annot_or,
 290           AlignmentAnnotation annot_new)
 291   {
 292     if (annot_or.annotations.length != annot_new.annotations.length)
 293     {
 294       System.err.println("Different lengths for annotation row elements: "+annot_or.annotations.length +"!="+ annot_new.annotations.length);
 295       return false;
 296     }
 297     for (int i = 0; i < annot_or.annotations.length; i++)
 298     {
 299       Annotation an_or=annot_or.annotations[i],an_new=annot_new.annotations[i];
 300       if (an_or != null
 301               && an_new!= null)
 302       {
 303         if (!an_or.displayCharacter.trim()
 304                 .equals(an_new.displayCharacter.trim())
 305                 || !(""+an_or.secondaryStructure).trim().equals((""+an_new.secondaryStructure).trim())
 306                 || (an_or.description != an_new.description && (an_or.description == null
 307                         || an_new.description == null || !an_or.description
 308                           .equals(an_new.description))))
 309         {
 310           System.err.println("Annotation Element Mismatch\nElement "+i+" in original: "+annot_or.annotations[i].toString()+"\nElement "+i+" in new: "+annot_new.annotations[i].toString());
 311           return false;
 312         }
 313       }
 314       else if (annot_or.annotations[i] == null
 315               && annot_new.annotations[i] == null)
 316       {
 317         continue;
 318       }
 319       else
 320       {
 321         System.err.println("Annotation Element Mismatch\nElement "+i+" in original: "+(annot_or.annotations[i]==null ? "is null" : annot_or.annotations[i].toString())+"\nElement "+i+" in new: "+(annot_new.annotations[i] == null ? "is null" : annot_new.annotations[i].toString()));
 322         return false;
 323       }
 324     }
 325     return true;
 326   }
 327 }