test/jalview/io/StockholmFileTest.java

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