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           assertTrue(
 151                   "Threshold line not identical.",
 152                   aa_original[i].threshold == null ? aa_new[i].threshold == null
 153                           : aa_original[i].threshold
 154                                   .equals(aa_new[i].threshold));
 155           // graphGroup may differ, but pattern should be the same
 156           Integer o_ggrp=new Integer(aa_original[i].graphGroup+2),n_ggrp=new Integer(aa_new[i].graphGroup+2);
 157           BitSet orig_g=orig_groups.get(o_ggrp),new_g=new_groups.get(n_ggrp);
 158           if (orig_g==null) {
 159             orig_groups.put(o_ggrp,orig_g= new BitSet());
 160           }
 161           if (new_g==null) {
 162             new_groups.put(n_ggrp, new_g=new BitSet());
 163           }
 164           assertTrue("Graph Group pattern differs at annotation "+i, orig_g.equals(new_g));
 165           orig_g.set(i); new_g.set(i);
 166         } else {
 167           System.err.println("No matching annotation row for "+aa_original[i].toString());
 168         }
 169       }
 170     }
 171     assertTrue(
 172             "Generated and imported alignment have different annotation sets ("
 173                     + aa_new_size + " != " + aa_original_size + ")",
 174             aa_new_size == aa_original_size);
 175
 176     // check sequences, annotation and features
 177     SequenceI[] seq_original = new SequenceI[al.getSequencesArray().length];
 178     seq_original = al.getSequencesArray();
 179     SequenceI[] seq_new = new SequenceI[al_input.getSequencesArray().length];
 180     seq_new = al_input.getSequencesArray();
 181     SequenceFeature[] sequenceFeatures_original, sequenceFeatures_new;
 182     AlignmentAnnotation annot_original, annot_new;
 183     //
 184     for (int i = 0; i < al.getSequencesArray().length; i++)
 185     {
 186       String name = seq_original[i].getName();
 187       int start = seq_original[i].getStart();
 188       int end = seq_original[i].getEnd();
 189       System.out.println("Check sequence: " + name + "/" + start + "-"
 190               + end);
 191
 192       // search equal sequence
 193       for (int in = 0; in < al_input.getSequencesArray().length; in++)
 194       {
 195         if (name.equals(seq_new[in].getName())
 196                 && start == seq_new[in].getStart()
 197                 && end == seq_new[in].getEnd())
 198         {
 199           String ss_original = seq_original[i].getSequenceAsString();
 200           String ss_new = seq_new[in].getSequenceAsString();
 201           assertTrue("The sequences " + name + "/" + start + "-" + end
 202                   + " are not equal", ss_original.equals(ss_new));
 203
 204           assertTrue(
 205                   "Sequence Features were not equivalent",
 206                   (seq_original[i].getSequenceFeatures() == null && seq_new[in]
 207                           .getSequenceFeatures() == null)
 208                           || (seq_original[i].getSequenceFeatures() != null && seq_new[in]
 209                                   .getSequenceFeatures() != null));
 210           // compare sequence features
 211           if (seq_original[i].getSequenceFeatures() != null
 212                   && seq_new[in].getSequenceFeatures() != null)
 213           {
 214             System.out.println("There are feature!!!");
 215             sequenceFeatures_original = new SequenceFeature[seq_original[i]
 216                     .getSequenceFeatures().length];
 217             sequenceFeatures_original = seq_original[i]
 218                     .getSequenceFeatures();
 219             sequenceFeatures_new = new SequenceFeature[seq_new[in]
 220                     .getSequenceFeatures().length];
 221             sequenceFeatures_new = seq_new[in].getSequenceFeatures();
 222
 223             assertTrue("different number of features", seq_original[i]
 224                     .getSequenceFeatures().length == seq_new[in]
 225                     .getSequenceFeatures().length);
 226
 227             for (int feat = 0; feat < seq_original[i].getSequenceFeatures().length; feat++)
 228             {
 229               assertTrue("Different features",
 230                       sequenceFeatures_original[feat]
 231                               .equals(sequenceFeatures_new[feat]));
 232             }
 233           }
 234           // compare alignment annotation
 235           if (al.getSequenceAt(i).getAnnotation() != null
 236                   && al_input.getSequenceAt(in).getAnnotation() != null)
 237           {
 238             for (int j = 0; j < al.getSequenceAt(i).getAnnotation().length; j++)
 239             {
 240               if (al.getSequenceAt(i).getAnnotation()[j] != null
 241                       && al_input.getSequenceAt(in).getAnnotation()[j] != null)
 242               {
 243                 annot_original = al.getSequenceAt(i).getAnnotation()[j];
 244                 annot_new = al_input.getSequenceAt(in).getAnnotation()[j];
 245                 assertTrue("Different annotation elements",
 246                         equalss(annot_original, annot_new));
 247               }
 248             }
 249           }
 250           else if (al.getSequenceAt(i).getAnnotation() == null
 251                   && al_input.getSequenceAt(in).getAnnotation() == null)
 252           {
 253             System.out.println("No annotations");
 254           }
 255           else if (al.getSequenceAt(i).getAnnotation() != null
 256                   && al_input.getSequenceAt(in).getAnnotation() == null)
 257           {
 258             assertTrue("Annotations differed between sequences ("
 259                     + al.getSequenceAt(i).getName() + ") and ("
 260                     + al_input.getSequenceAt(i).getName() + ")", false);
 261           }
 262           break;
 263         }
 264       }
 265     }
 266   }
 267
 268   /*
 269    * compare annotations
 270    */
 271   private static boolean equalss(AlignmentAnnotation annot_or,
 272           AlignmentAnnotation annot_new)
 273   {
 274     if (annot_or.annotations.length != annot_new.annotations.length)
 275     {
 276       System.err.println("Different lengths for annotation row elements: "+annot_or.annotations.length +"!="+ annot_new.annotations.length);
 277       return false;
 278     }
 279     for (int i = 0; i < annot_or.annotations.length; i++)
 280     {
 281       Annotation an_or=annot_or.annotations[i],an_new=annot_new.annotations[i];
 282       if (an_or != null
 283               && an_new!= null)
 284       {
 285         if (!an_or.displayCharacter.trim()
 286                 .equals(an_new.displayCharacter.trim())
 287                 || !(""+an_or.secondaryStructure).trim().equals((""+an_new.secondaryStructure).trim())
 288                 || (an_or.description != an_new.description && (an_or.description == null
 289                         || an_new.description == null || !an_or.description
 290                           .equals(an_new.description))))
 291         {
 292           System.err.println("Annotation Element Mismatch\nElement "+i+" in original: "+annot_or.annotations[i].toString()+"\nElement "+i+" in new: "+annot_new.annotations[i].toString());
 293           return false;
 294         }
 295       }
 296       else if (annot_or.annotations[i] == null
 297               && annot_new.annotations[i] == null)
 298       {
 299         continue;
 300       }
 301       else
 302       {
 303         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()));
 304         return false;
 305       }
 306     }
 307     return true;
 308   }
 309 }