test/jalview/io/StockholmFileTest.java

   1 /*
   2  * Jalview - A Sequence Alignment Editor and Viewer (Version 2.8.2)
   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
  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.assertNotNull;
  24 import static org.junit.Assert.assertTrue;
  25 import jalview.datamodel.Alignment;
  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");
  49   }
  50   @Test
  51   public void pfamFileDataExtraction() throws Exception
  52   {  AppletFormatAdapter af = new AppletFormatAdapter();
  53     AlignmentI al = af.readFile(PfamFile, af.FILE,
  54             new IdentifyFile().Identify(PfamFile, af.FILE));
  55     int numpdb = 0;
  56     for (SequenceI sq : al.getSequences())
  57     {
  58       if (sq.getPDBId() != null)
  59       {
  60         numpdb += sq.getPDBId().size();
  61       }
  62     }
  63     assertTrue(
  64             "PF00111 seed alignment has at least 1 PDB file, but the reader found none.",
  65             numpdb > 0);
  66   }
  67
  68   @Test
  69   public void rfamFileIO() throws Exception
  70   {
  71     testFileIOwithFormat(new File(RfamFile), "STH");
  72   }
  73
  74   /**
  75    * test alignment data in given file can be imported, exported and reimported
  76    * with no dataloss
  77    *
  78    * @param f
  79    *          - source datafile (IdentifyFile.identify() should work with it)
  80    * @param ioformat
  81    *          - label for IO class used to write and read back in the data from
  82    *          f
  83    */
  84   public static void testFileIOwithFormat(File f, String ioformat)
  85   {
  86     System.out.println("Reading file: " + f);
  87     String ff = f.getPath();
  88     try
  89     {
  90       AppletFormatAdapter rf = new AppletFormatAdapter();
  91
  92       Alignment al = rf.readFile(ff, AppletFormatAdapter.FILE,
  93               new IdentifyFile().Identify(ff, AppletFormatAdapter.FILE));
  94
  95       assertNotNull("Couldn't read supplied alignment data.", al);
  96
  97       // make sure dataset is initialised ? not sure about this
  98       for (int i = 0; i < al.getSequencesArray().length; ++i)
  99       {
 100         al.getSequenceAt(i).setDatasetSequence(al.getSequenceAt(i));
 101       }
 102       String outputfile = rf.formatSequences(ioformat, al, true);
 103       System.out.println("Output file in '"+ioformat+"':\n"+outputfile+"\n<<EOF\n");
 104       // test for consistency in io
 105       Alignment al_input = new AppletFormatAdapter().readFile(outputfile,
 106               AppletFormatAdapter.PASTE, ioformat);
 107       assertNotNull("Couldn't parse reimported alignment data.", al_input);
 108
 109       String identifyoutput = new IdentifyFile().Identify(outputfile,
 110               AppletFormatAdapter.PASTE);
 111       assertNotNull("Identify routine failed for outputformat " + ioformat,
 112               identifyoutput);
 113       assertTrue(
 114               "Identify routine could not recognise output generated by '"
 115                       + ioformat + "' writer",
 116               ioformat.equals(identifyoutput));
 117       testAlignmentEquivalence(al, al_input);
 118     } catch (Exception e)
 119     {
 120       e.printStackTrace();
 121       assertTrue("Couln't format the alignment for output file.", false);
 122     }
 123   }
 124
 125   /**
 126    * assert alignment equivalence
 127    *
 128    * @param al
 129    *          'original'
 130    * @param al_input
 131    *          'secondary' or generated alignment from some datapreserving
 132    *          transformation
 133    */
 134   public static void testAlignmentEquivalence(AlignmentI al,
 135           AlignmentI al_input)
 136   {
 137     assertNotNull("Original alignment was null", al);
 138     assertNotNull("Generated alignment was null", al_input);
 139
 140     assertTrue(
 141             "Alignment dimension mismatch: originl contains "
 142                     + al.getHeight() + " and generated has "
 143                     + al_input.getHeight() + " sequences; original has "
 144                     + al.getWidth() + " and generated has "
 145                     + al_input.getWidth() + " columns.",
 146             al.getHeight() == al_input.getHeight()
 147                     && al.getWidth() == al_input.getWidth());
 148
 149     // check Alignment annotation
 150     AlignmentAnnotation[] aa_new = al_input.getAlignmentAnnotation();
 151     AlignmentAnnotation[] aa_original = al.getAlignmentAnnotation();
 152
 153     // note - at moment we do not distinguish between alignment without any
 154     // annotation rows and alignment with no annotation row vector
 155     // we might want to revise this in future
 156     int aa_new_size = (aa_new == null ? 0 : aa_new.length), aa_original_size = (aa_original == null ? 0
 157             : aa_original.length);
 158     Map<Integer,java.util.BitSet> orig_groups=new HashMap<Integer,java.util.BitSet>(),new_groups=new HashMap<Integer,java.util.BitSet>();
 159
 160     if (aa_new != null && aa_original != null)
 161     {
 162       for (int i = 0; i < aa_original.length; i++)
 163       {
 164         if (aa_new.length>i) {
 165           assertTrue("Different alignment annotation at position "+i,
 166                 equalss(aa_original[i], aa_new[i]));
 167           // compare graphGroup or graph properties - needed to verify JAL-1299
 168           assertTrue("Graph type not identical.",aa_original[i].graph==aa_new[i].graph);
 169           assertTrue("Visibility not identical.", aa_original[i].visible==aa_new[i].visible);
 170           assertTrue(
 171                   "Threshold line not identical.",
 172                   aa_original[i].threshold == null ? aa_new[i].threshold == null
 173                           : aa_original[i].threshold
 174                                   .equals(aa_new[i].threshold));
 175           // graphGroup may differ, but pattern should be the same
 176           Integer o_ggrp=new Integer(aa_original[i].graphGroup+2),n_ggrp=new Integer(aa_new[i].graphGroup+2);
 177           BitSet orig_g=orig_groups.get(o_ggrp),new_g=new_groups.get(n_ggrp);
 178           if (orig_g==null) {
 179             orig_groups.put(o_ggrp,orig_g= new BitSet());
 180           }
 181           if (new_g==null) {
 182             new_groups.put(n_ggrp, new_g=new BitSet());
 183           }
 184           assertTrue("Graph Group pattern differs at annotation "+i, orig_g.equals(new_g));
 185           orig_g.set(i); new_g.set(i);
 186         } else {
 187           System.err.println("No matching annotation row for "+aa_original[i].toString());
 188         }
 189       }
 190     }
 191     assertTrue(
 192             "Generated and imported alignment have different annotation sets ("
 193                     + aa_new_size + " != " + aa_original_size + ")",
 194             aa_new_size == aa_original_size);
 195
 196     // check sequences, annotation and features
 197     SequenceI[] seq_original = new SequenceI[al.getSequencesArray().length];
 198     seq_original = al.getSequencesArray();
 199     SequenceI[] seq_new = new SequenceI[al_input.getSequencesArray().length];
 200     seq_new = al_input.getSequencesArray();
 201     SequenceFeature[] sequenceFeatures_original, sequenceFeatures_new;
 202     AlignmentAnnotation annot_original, annot_new;
 203     //
 204     for (int i = 0; i < al.getSequencesArray().length; i++)
 205     {
 206       String name = seq_original[i].getName();
 207       int start = seq_original[i].getStart();
 208       int end = seq_original[i].getEnd();
 209       System.out.println("Check sequence: " + name + "/" + start + "-"
 210               + end);
 211
 212       // search equal sequence
 213       for (int in = 0; in < al_input.getSequencesArray().length; in++)
 214       {
 215         if (name.equals(seq_new[in].getName())
 216                 && start == seq_new[in].getStart()
 217                 && end == seq_new[in].getEnd())
 218         {
 219           String ss_original = seq_original[i].getSequenceAsString();
 220           String ss_new = seq_new[in].getSequenceAsString();
 221           assertTrue("The sequences " + name + "/" + start + "-" + end
 222                   + " are not equal", ss_original.equals(ss_new));
 223
 224           assertTrue(
 225                   "Sequence Features were not equivalent",
 226                   (seq_original[i].getSequenceFeatures() == null && seq_new[in]
 227                           .getSequenceFeatures() == null)
 228                           || (seq_original[i].getSequenceFeatures() != null && seq_new[in]
 229                                   .getSequenceFeatures() != null));
 230           // compare sequence features
 231           if (seq_original[i].getSequenceFeatures() != null
 232                   && seq_new[in].getSequenceFeatures() != null)
 233           {
 234             System.out.println("There are feature!!!");
 235             sequenceFeatures_original = new SequenceFeature[seq_original[i]
 236                     .getSequenceFeatures().length];
 237             sequenceFeatures_original = seq_original[i]
 238                     .getSequenceFeatures();
 239             sequenceFeatures_new = new SequenceFeature[seq_new[in]
 240                     .getSequenceFeatures().length];
 241             sequenceFeatures_new = seq_new[in].getSequenceFeatures();
 242
 243             assertTrue("different number of features", seq_original[i]
 244                     .getSequenceFeatures().length == seq_new[in]
 245                     .getSequenceFeatures().length);
 246
 247             for (int feat = 0; feat < seq_original[i].getSequenceFeatures().length; feat++)
 248             {
 249               assertTrue("Different features",
 250                       sequenceFeatures_original[feat]
 251                               .equals(sequenceFeatures_new[feat]));
 252             }
 253           }
 254           // compare alignment annotation
 255           if (al.getSequenceAt(i).getAnnotation() != null
 256                   && al_input.getSequenceAt(in).getAnnotation() != null)
 257           {
 258             for (int j = 0; j < al.getSequenceAt(i).getAnnotation().length; j++)
 259             {
 260               if (al.getSequenceAt(i).getAnnotation()[j] != null
 261                       && al_input.getSequenceAt(in).getAnnotation()[j] != null)
 262               {
 263                 annot_original = al.getSequenceAt(i).getAnnotation()[j];
 264                 annot_new = al_input.getSequenceAt(in).getAnnotation()[j];
 265                 assertTrue("Different annotation elements",
 266                         equalss(annot_original, annot_new));
 267               }
 268             }
 269           }
 270           else if (al.getSequenceAt(i).getAnnotation() == null
 271                   && al_input.getSequenceAt(in).getAnnotation() == null)
 272           {
 273             System.out.println("No annotations");
 274           }
 275           else if (al.getSequenceAt(i).getAnnotation() != null
 276                   && al_input.getSequenceAt(in).getAnnotation() == null)
 277           {
 278             assertTrue("Annotations differed between sequences ("
 279                     + al.getSequenceAt(i).getName() + ") and ("
 280                     + al_input.getSequenceAt(i).getName() + ")", false);
 281           }
 282           break;
 283         }
 284       }
 285     }
 286   }
 287
 288   /*
 289    * compare annotations
 290    */
 291   private static boolean equalss(AlignmentAnnotation annot_or,
 292           AlignmentAnnotation annot_new)
 293   {
 294     if (annot_or.annotations.length != annot_new.annotations.length)
 295     {
 296       System.err.println("Different lengths for annotation row elements: "+annot_or.annotations.length +"!="+ annot_new.annotations.length);
 297       return false;
 298     }
 299     for (int i = 0; i < annot_or.annotations.length; i++)
 300     {
 301       Annotation an_or=annot_or.annotations[i],an_new=annot_new.annotations[i];
 302       if (an_or != null
 303               && an_new!= null)
 304       {
 305         if (!an_or.displayCharacter.trim()
 306                 .equals(an_new.displayCharacter.trim())
 307                 || !(""+an_or.secondaryStructure).trim().equals((""+an_new.secondaryStructure).trim())
 308                 || (an_or.description != an_new.description && (an_or.description == null
 309                         || an_new.description == null || !an_or.description
 310                           .equals(an_new.description))))
 311         {
 312           System.err.println("Annotation Element Mismatch\nElement "+i+" in original: "+annot_or.annotations[i].toString()+"\nElement "+i+" in new: "+annot_new.annotations[i].toString());
 313           return false;
 314         }
 315       }
 316       else if (annot_or.annotations[i] == null
 317               && annot_new.annotations[i] == null)
 318       {
 319         continue;
 320       }
 321       else
 322       {
 323         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()));
 324         return false;
 325       }
 326     }
 327     return true;
 328   }
 329 }