forester/java/src/org/forester/tools/PhylogenyDecorator.java

   1 // $Id:
   2 // FORESTER -- software libraries and applications
   3 // for evolutionary biology research and applications.
   4 //
   5 // Copyright (C) 2008-2009 Christian M. Zmasek
   6 // Copyright (C) 2008-2009 Burnham Institute for Medical Research
   7 // All rights reserved
   8 //
   9 // This library is free software; you can redistribute it and/or
  10 // modify it under the terms of the GNU Lesser General Public
  11 // License as published by the Free Software Foundation; either
  12 // version 2.1 of the License, or (at your option) any later version.
  13 //
  14 // This library is distributed in the hope that it will be useful,
  15 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  16 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  17 // Lesser General Public License for more details.
  18 //
  19 // You should have received a copy of the GNU Lesser General Public
  20 // License along with this library; if not, write to the Free Software
  21 // Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
  22 //
  23 // Contact: phylosoft @ gmail . com
  24 // WWW: www.phylosoft.org/forester
  25
  26 package org.forester.tools;
  27
  28 import java.io.File;
  29 import java.io.IOException;
  30 import java.util.HashMap;
  31 import java.util.Map;
  32 import java.util.regex.Matcher;
  33 import java.util.regex.Pattern;
  34
  35 import org.forester.archaeopteryx.AptxUtil;
  36 import org.forester.io.parsers.nhx.NHXFormatException;
  37 import org.forester.io.parsers.phyloxml.PhyloXmlDataFormatException;
  38 import org.forester.phylogeny.Phylogeny;
  39 import org.forester.phylogeny.PhylogenyNode;
  40 import org.forester.phylogeny.data.Accession;
  41 import org.forester.phylogeny.data.Annotation;
  42 import org.forester.phylogeny.data.DomainArchitecture;
  43 import org.forester.phylogeny.data.Identifier;
  44 import org.forester.phylogeny.data.Sequence;
  45 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
  46 import org.forester.util.BasicTable;
  47 import org.forester.util.BasicTableParser;
  48 import org.forester.util.ForesterUtil;
  49
  50 public final class PhylogenyDecorator {
  51
  52     // From evoruby/lib/evo/apps/tseq_taxonomy_processor.rb:
  53     final private static String  TP_TAXONOMY_CODE                   = "TAXONOMY_CODE";
  54     final private static String  TP_TAXONOMY_ID                     = "TAXONOMY_ID";
  55     final private static String  TP_TAXONOMY_ID_PROVIDER            = "TAXONOMY_ID_PROVIDER";
  56     final private static String  TP_TAXONOMY_SN                     = "TAXONOMY_SN";
  57     final private static String  TP_TAXONOMY_CN                     = "TAXONOMY_CN";
  58     final private static String  TP_TAXONOMY_SYN                    = "TAXONOMY_SYN";
  59     final private static String  TP_SEQ_SYMBOL                      = "SEQ_SYMBOL";
  60     final private static String  TP_SEQ_ACCESSION                   = "SEQ_ACCESSION";
  61     final private static String  TP_SEQ_ACCESSION_SOURCE            = "SEQ_ACCESSION_SOURCE";
  62     final private static String  TP_SEQ_ANNOTATION_DESC             = "SEQ_ANNOTATION_DESC";
  63     final private static String  TP_SEQ_ANNOTATION_REF              = "SEQ_ANNOTATION_REF";
  64     final private static String  TP_SEQ_MOL_SEQ                     = "SEQ_MOL_SEQ";
  65     final private static String  TP_SEQ_NAME                        = "SEQ_NAME";
  66     final private static String  TP_NODE_NAME                       = "NODE_NAME";
  67     final private static Pattern NODENAME_SEQNUMBER_TAXDOMAINNUMBER = Pattern
  68                                                                             .compile( "^([a-fA-Z0-9]{1,5})_([A-Z0-9]{2,4}[A-Z])(\\d{1,4})$" );
  69     public final static boolean  SANITIZE                           = false;
  70     public final static boolean  VERBOSE                            = true;
  71
  72     private PhylogenyDecorator() {
  73         // Not needed.
  74     }
  75
  76     public static void decorate( final Phylogeny phylogeny,
  77                                  final Map<String, Map<String, String>> map,
  78                                  final boolean picky,
  79                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map )
  80             throws IllegalArgumentException, PhyloXmlDataFormatException {
  81         for( final PhylogenyNodeIterator iter = phylogeny.iteratorPostorder(); iter.hasNext(); ) {
  82             final PhylogenyNode node = iter.next();
  83             final String name = node.getName();
  84             if ( !ForesterUtil.isEmpty( name ) ) {
  85                 if ( map.containsKey( name ) || ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 ) ) {
  86                     Map<String, String> new_values = map.get( name );
  87                     int x = 0;
  88                     while ( ( new_values == null ) && ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 )
  89                             && ( x <= numbers_of_chars_allowed_to_remove_if_not_found_in_map ) ) {
  90                         new_values = map.get( name.substring( 0, name.length() - x ) );
  91                         ++x;
  92                     }
  93                     if ( new_values != null ) {
  94                         if ( new_values.containsKey( TP_TAXONOMY_CODE ) ) {
  95                             AptxUtil.ensurePresenceOfTaxonomy( node );
  96                             node.getNodeData().getTaxonomy().setTaxonomyCode( new_values.get( TP_TAXONOMY_CODE ) );
  97                         }
  98                         if ( new_values.containsKey( TP_TAXONOMY_ID )
  99                                 && new_values.containsKey( TP_TAXONOMY_ID_PROVIDER ) ) {
 100                             AptxUtil.ensurePresenceOfTaxonomy( node );
 101                             node.getNodeData()
 102                                     .getTaxonomy()
 103                                     .setIdentifier( new Identifier( new_values.get( TP_TAXONOMY_ID ),
 104                                                                     new_values.get( TP_TAXONOMY_ID_PROVIDER ) ) );
 105                         }
 106                         else if ( new_values.containsKey( TP_TAXONOMY_ID ) ) {
 107                             AptxUtil.ensurePresenceOfTaxonomy( node );
 108                             node.getNodeData().getTaxonomy()
 109                                     .setIdentifier( new Identifier( new_values.get( TP_TAXONOMY_ID ) ) );
 110                         }
 111                         if ( new_values.containsKey( TP_TAXONOMY_SN ) ) {
 112                             AptxUtil.ensurePresenceOfTaxonomy( node );
 113                             node.getNodeData().getTaxonomy().setScientificName( new_values.get( TP_TAXONOMY_SN ) );
 114                         }
 115                         if ( new_values.containsKey( TP_TAXONOMY_CN ) ) {
 116                             AptxUtil.ensurePresenceOfTaxonomy( node );
 117                             node.getNodeData().getTaxonomy().setCommonName( new_values.get( TP_TAXONOMY_CN ) );
 118                         }
 119                         if ( new_values.containsKey( TP_TAXONOMY_SYN ) ) {
 120                             AptxUtil.ensurePresenceOfTaxonomy( node );
 121                             node.getNodeData().getTaxonomy().getSynonyms().add( new_values.get( TP_TAXONOMY_SYN ) );
 122                         }
 123                         if ( new_values.containsKey( TP_SEQ_ACCESSION )
 124                                 && new_values.containsKey( TP_SEQ_ACCESSION_SOURCE ) ) {
 125                             AptxUtil.ensurePresenceOfSequence( node );
 126                             node.getNodeData()
 127                                     .getSequence()
 128                                     .setAccession( new Accession( new_values.get( TP_SEQ_ACCESSION ),
 129                                                                   new_values.get( TP_SEQ_ACCESSION_SOURCE ) ) );
 130                         }
 131                         if ( new_values.containsKey( TP_SEQ_ANNOTATION_DESC ) ) {
 132                             AptxUtil.ensurePresenceOfSequence( node );
 133                             final Annotation ann = new Annotation( "?" );
 134                             ann.setDesc( new_values.get( TP_SEQ_ANNOTATION_DESC ) );
 135                             node.getNodeData().getSequence().addAnnotation( ann );
 136                         }
 137                         if ( new_values.containsKey( TP_SEQ_ANNOTATION_REF ) ) {
 138                             AptxUtil.ensurePresenceOfSequence( node );
 139                             final Annotation ann = new Annotation( new_values.get( TP_SEQ_ANNOTATION_REF ) );
 140                             node.getNodeData().getSequence().addAnnotation( ann );
 141                         }
 142                         if ( new_values.containsKey( TP_SEQ_SYMBOL ) ) {
 143                             AptxUtil.ensurePresenceOfSequence( node );
 144                             node.getNodeData().getSequence().setSymbol( new_values.get( TP_SEQ_SYMBOL ) );
 145                         }
 146                         if ( new_values.containsKey( TP_SEQ_NAME ) ) {
 147                             AptxUtil.ensurePresenceOfSequence( node );
 148                             node.getNodeData().getSequence().setName( new_values.get( TP_SEQ_NAME ) );
 149                         }
 150                         if ( new_values.containsKey( TP_SEQ_MOL_SEQ ) ) {
 151                             AptxUtil.ensurePresenceOfSequence( node );
 152                             node.getNodeData().getSequence().setMolecularSequence( new_values.get( TP_SEQ_MOL_SEQ ) );
 153                         }
 154                         if ( new_values.containsKey( TP_NODE_NAME ) ) {
 155                             node.setName( new_values.get( TP_NODE_NAME ) );
 156                         }
 157                     } // if ( new_values != null )
 158                 } // if ( map.containsKey( name ) || ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 ) )
 159                 else if ( picky ) {
 160                     throw new IllegalArgumentException( "\"" + name + "\" not found in name map" );
 161                 }
 162             }
 163         }
 164     }
 165
 166     /**
 167      *
 168      *
 169      *
 170      *
 171      *
 172      * @param phylogeny
 173      * @param map
 174      *            maps names (in phylogeny) to new values
 175      * @param field
 176      * @param picky
 177      * @throws IllegalArgumentException
 178      * @throws NHXFormatException
 179      * @throws PhyloXmlDataFormatException
 180      */
 181     public static void decorate( final Phylogeny phylogeny,
 182                                  final Map<String, String> map,
 183                                  final FIELD field,
 184                                  final boolean extract_bracketed_scientific_name,
 185                                  final boolean picky,
 186                                  final boolean cut_name_after_space,
 187                                  final boolean process_name_intelligently,
 188                                  final boolean process_similar_to,
 189                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 190                                  final boolean move_domain_numbers_at_end_to_middle ) throws IllegalArgumentException,
 191             NHXFormatException, PhyloXmlDataFormatException {
 192         PhylogenyDecorator.decorate( phylogeny,
 193                                      map,
 194                                      field,
 195                                      extract_bracketed_scientific_name,
 196                                      picky,
 197                                      null,
 198                                      cut_name_after_space,
 199                                      process_name_intelligently,
 200                                      process_similar_to,
 201                                      numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 202                                      move_domain_numbers_at_end_to_middle );
 203     }
 204
 205     /**
 206      *
 207      *
 208      *
 209      * @param phylogeny
 210      * @param map
 211      *            maps names (in phylogeny) to new values if intermediate_map is
 212      *            null otherwise maps intermediate value to new value
 213      * @param field
 214      * @param picky
 215      * @param intermediate_map
 216      *            maps name (in phylogeny) to a intermediate value
 217      * @throws IllegalArgumentException
 218      * @throws PhyloXmlDataFormatException
 219      */
 220     public static void decorate( final Phylogeny phylogeny,
 221                                  final Map<String, String> map,
 222                                  final FIELD field,
 223                                  final boolean extract_bracketed_scientific_name,
 224                                  final boolean picky,
 225                                  final Map<String, String> intermediate_map,
 226                                  final boolean cut_name_after_space,
 227                                  final boolean process_name_intelligently,
 228                                  final boolean process_similar_to,
 229                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 230                                  final boolean move_domain_numbers_at_end_to_middle ) throws IllegalArgumentException,
 231             PhyloXmlDataFormatException {
 232         if ( extract_bracketed_scientific_name && ( field == FIELD.TAXONOMY_SCIENTIFIC_NAME ) ) {
 233             throw new IllegalArgumentException( "Attempt to extract bracketed scientific name together with data field pointing to scientific name" );
 234         }
 235         for( final PhylogenyNodeIterator iter = phylogeny.iteratorPostorder(); iter.hasNext(); ) {
 236             final PhylogenyNode node = iter.next();
 237             String name = node.getName();
 238             if ( !ForesterUtil.isEmpty( name ) ) {
 239                 if ( intermediate_map != null ) {
 240                     name = PhylogenyDecorator.extractIntermediate( intermediate_map, name );
 241                 }
 242                 if ( map.containsKey( name ) || ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 ) ) {
 243                     String new_value = map.get( name );
 244                     int x = 0;
 245                     while ( ( new_value == null ) && ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 )
 246                             && ( x <= numbers_of_chars_allowed_to_remove_if_not_found_in_map ) ) {
 247                         new_value = map.get( name.substring( 0, name.length() - x ) );
 248                         ++x;
 249                     }
 250                     if ( new_value != null ) {
 251                         new_value = new_value.trim();
 252                         new_value.replaceAll( "/\\s+/", " " );
 253                         if ( extract_bracketed_scientific_name && new_value.endsWith( "]" ) ) {
 254                             extractBracketedScientificNames( node, new_value );
 255                         }
 256                         switch ( field ) {
 257                             case SEQUENCE_ANNOTATION_DESC:
 258                                 if ( PhylogenyDecorator.VERBOSE ) {
 259                                     System.out.println( name + ": " + new_value );
 260                                 }
 261                                 if ( !node.getNodeData().isHasSequence() ) {
 262                                     node.getNodeData().setSequence( new Sequence() );
 263                                 }
 264                                 final Annotation annotation = new Annotation( "?" );
 265                                 annotation.setDesc( new_value );
 266                                 node.getNodeData().getSequence().addAnnotation( annotation );
 267                                 break;
 268                             case DOMAIN_STRUCTURE:
 269                                 if ( PhylogenyDecorator.VERBOSE ) {
 270                                     System.out.println( name + ": " + new_value );
 271                                 }
 272                                 if ( !node.getNodeData().isHasSequence() ) {
 273                                     node.getNodeData().setSequence( new Sequence() );
 274                                 }
 275                                 node.getNodeData().getSequence()
 276                                         .setDomainArchitecture( new DomainArchitecture( new_value ) );
 277                                 break;
 278                             case TAXONOMY_CODE:
 279                                 if ( PhylogenyDecorator.VERBOSE ) {
 280                                     System.out.println( name + ": " + new_value );
 281                                 }
 282                                 AptxUtil.ensurePresenceOfTaxonomy( node );
 283                                 node.getNodeData().getTaxonomy().setTaxonomyCode( new_value );
 284                                 break;
 285                             case TAXONOMY_SCIENTIFIC_NAME:
 286                                 if ( PhylogenyDecorator.VERBOSE ) {
 287                                     System.out.println( name + ": " + new_value );
 288                                 }
 289                                 AptxUtil.ensurePresenceOfTaxonomy( node );
 290                                 node.getNodeData().getTaxonomy().setScientificName( new_value );
 291                                 break;
 292                             case SEQUENCE_NAME:
 293                                 if ( PhylogenyDecorator.VERBOSE ) {
 294                                     System.out.println( name + ": " + new_value );
 295                                 }
 296                                 if ( !node.getNodeData().isHasSequence() ) {
 297                                     node.getNodeData().setSequence( new Sequence() );
 298                                 }
 299                                 node.getNodeData().getSequence().setName( new_value );
 300                                 break;
 301                             case NODE_NAME:
 302                                 if ( PhylogenyDecorator.VERBOSE ) {
 303                                     System.out.print( name + " -> " );
 304                                 }
 305                                 if ( cut_name_after_space ) {
 306                                     if ( PhylogenyDecorator.VERBOSE ) {
 307                                         System.out.print( new_value + " -> " );
 308                                     }
 309                                     new_value = PhylogenyDecorator.deleteAtFirstSpace( new_value );
 310                                 }
 311                                 else if ( process_name_intelligently ) {
 312                                     if ( PhylogenyDecorator.VERBOSE ) {
 313                                         System.out.print( new_value + " -> " );
 314                                     }
 315                                     new_value = PhylogenyDecorator.processNameIntelligently( new_value );
 316                                 }
 317                                 else if ( process_similar_to ) {
 318                                     if ( PhylogenyDecorator.VERBOSE ) {
 319                                         System.out.print( new_value + " -> " );
 320                                     }
 321                                     new_value = PhylogenyDecorator.processSimilarTo( new_value );
 322                                 }
 323                                 if ( PhylogenyDecorator.SANITIZE ) {
 324                                     new_value = PhylogenyDecorator.sanitize( new_value );
 325                                 }
 326                                 if ( PhylogenyDecorator.VERBOSE ) {
 327                                     System.out.println( new_value );
 328                                 }
 329                                 node.setName( new_value );
 330                                 break;
 331                             default:
 332                                 throw new RuntimeException( "unknown field \"" + field + "\"" );
 333                         }
 334                         if ( move_domain_numbers_at_end_to_middle && ( field != FIELD.NODE_NAME ) ) {
 335                             node.setName( moveDomainNumbersAtEnd( node.getName() ) );
 336                         }
 337                     }
 338                 }
 339                 else if ( picky ) {
 340                     throw new IllegalArgumentException( "\"" + name + "\" not found in name map" );
 341                 }
 342             }
 343         }
 344     }
 345
 346     public static void decorate( final Phylogeny[] phylogenies,
 347                                  final Map<String, Map<String, String>> map,
 348                                  final boolean picky,
 349                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map )
 350             throws IllegalArgumentException, NHXFormatException, PhyloXmlDataFormatException {
 351         for( int i = 0; i < phylogenies.length; ++i ) {
 352             PhylogenyDecorator.decorate( phylogenies[ i ],
 353                                          map,
 354                                          picky,
 355                                          numbers_of_chars_allowed_to_remove_if_not_found_in_map );
 356         }
 357     }
 358
 359     public static void decorate( final Phylogeny[] phylogenies,
 360                                  final Map<String, String> map,
 361                                  final FIELD field,
 362                                  final boolean extract_bracketed_scientific_name,
 363                                  final boolean picky,
 364                                  final boolean cut_name_after_space,
 365                                  final boolean process_name_intelligently,
 366                                  final boolean process_similar_to,
 367                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 368                                  final boolean move_domain_numbers_at_end_to_middle ) throws IllegalArgumentException,
 369             NHXFormatException, PhyloXmlDataFormatException {
 370         for( int i = 0; i < phylogenies.length; ++i ) {
 371             PhylogenyDecorator.decorate( phylogenies[ i ],
 372                                          map,
 373                                          field,
 374                                          extract_bracketed_scientific_name,
 375                                          picky,
 376                                          cut_name_after_space,
 377                                          process_name_intelligently,
 378                                          process_similar_to,
 379                                          numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 380                                          move_domain_numbers_at_end_to_middle );
 381         }
 382     }
 383
 384     public static void decorate( final Phylogeny[] phylogenies,
 385                                  final Map<String, String> map,
 386                                  final FIELD field,
 387                                  final boolean extract_bracketed_scientific_name,
 388                                  final boolean picky,
 389                                  final Map<String, String> intermediate_map,
 390                                  final boolean cut_name_after_space,
 391                                  final boolean process_name_intelligently,
 392                                  final boolean process_similar_to,
 393                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 394                                  final boolean move_domain_numbers_at_end_to_middle ) throws IllegalArgumentException,
 395             NHXFormatException, PhyloXmlDataFormatException {
 396         for( int i = 0; i < phylogenies.length; ++i ) {
 397             PhylogenyDecorator.decorate( phylogenies[ i ],
 398                                          map,
 399                                          field,
 400                                          extract_bracketed_scientific_name,
 401                                          picky,
 402                                          intermediate_map,
 403                                          cut_name_after_space,
 404                                          process_name_intelligently,
 405                                          process_similar_to,
 406                                          numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 407                                          move_domain_numbers_at_end_to_middle );
 408         }
 409     }
 410
 411     private static String deleteAtFirstSpace( final String name ) {
 412         final int first_space = name.indexOf( " " );
 413         if ( first_space > 1 ) {
 414             return name.substring( 0, first_space ).trim();
 415         }
 416         return name;
 417     }
 418
 419     private static void extractBracketedScientificNames( final PhylogenyNode node, final String new_value ) {
 420         final int i = new_value.lastIndexOf( "[" );
 421         final String scientific_name = new_value.substring( i + 1, new_value.length() - 1 );
 422         AptxUtil.ensurePresenceOfTaxonomy( node );
 423         node.getNodeData().getTaxonomy().setScientificName( scientific_name );
 424     }
 425
 426     private static String extractIntermediate( final Map<String, String> intermediate_map, final String name ) {
 427         String new_name = null;
 428         if ( PhylogenyDecorator.VERBOSE ) {
 429             System.out.print( name + " => " );
 430         }
 431         if ( intermediate_map.containsKey( name ) ) {
 432             new_name = intermediate_map.get( name );
 433             if ( ForesterUtil.isEmpty( new_name ) ) {
 434                 throw new IllegalArgumentException( "\"" + name + "\" maps to null or empty string in secondary map" );
 435             }
 436         }
 437         else {
 438             throw new IllegalArgumentException( "\"" + name + "\" not found in name secondary map" );
 439         }
 440         if ( PhylogenyDecorator.VERBOSE ) {
 441             System.out.println( new_name + "  " );
 442         }
 443         return new_name;
 444     }
 445
 446     private static String moveDomainNumbersAtEnd( final String node_name ) {
 447         final Matcher m = NODENAME_SEQNUMBER_TAXDOMAINNUMBER.matcher( node_name );
 448         if ( m.matches() ) {
 449             final String seq_number = m.group( 1 );
 450             final String tax = m.group( 2 );
 451             final String domain_number = m.group( 3 );
 452             return seq_number + "_[" + domain_number + "]_" + tax;
 453         }
 454         else {
 455             return node_name;
 456         }
 457     }
 458
 459     public static Map<String, Map<String, String>> parseMappingTable( final File mapping_table_file )
 460             throws IOException {
 461         final Map<String, Map<String, String>> map = new HashMap<String, Map<String, String>>();
 462         BasicTable<String> mapping_table = null;
 463         mapping_table = BasicTableParser.parse( mapping_table_file, "\t", false );
 464         for( int row = 0; row < mapping_table.getNumberOfRows(); ++row ) {
 465             final Map<String, String> row_map = new HashMap<String, String>();
 466             String name = null;
 467             for( int col = 0; col < mapping_table.getNumberOfColumns(); ++col ) {
 468                 final String table_cell = mapping_table.getValue( col, row );
 469                 if ( col == 0 ) {
 470                     name = table_cell;
 471                 }
 472                 else if ( table_cell != null ) {
 473                     final String key = table_cell.substring( 0, table_cell.indexOf( ':' ) );
 474                     final String val = table_cell.substring( table_cell.indexOf( ':' ) + 1, table_cell.length() );
 475                     row_map.put( key, val );
 476                 }
 477             }
 478             map.put( name, row_map );
 479         }
 480         return map;
 481     }
 482
 483     private static String processNameIntelligently( final String name ) {
 484         final String[] s = name.split( " " );
 485         if ( s.length < 2 ) {
 486             return name;
 487         }
 488         else if ( ( s[ 0 ].indexOf( "_" ) > 0 ) && ( s[ 0 ].indexOf( "|" ) > 0 ) ) {
 489             return s[ 0 ];
 490         }
 491         else if ( ( s[ 1 ].indexOf( "_" ) > 0 ) && ( s[ 1 ].indexOf( "|" ) > 0 ) ) {
 492             return s[ 1 ];
 493         }
 494         else if ( ( s[ 0 ].indexOf( "_" ) > 0 ) && ( s[ 0 ].indexOf( "." ) > 0 ) ) {
 495             return s[ 0 ];
 496         }
 497         else if ( ( s[ 1 ].indexOf( "_" ) > 0 ) && ( s[ 1 ].indexOf( "." ) > 0 ) ) {
 498             return s[ 1 ];
 499         }
 500         else if ( s[ 0 ].indexOf( "_" ) > 0 ) {
 501             return s[ 0 ];
 502         }
 503         else if ( s[ 1 ].indexOf( "_" ) > 0 ) {
 504             return s[ 1 ];
 505         }
 506         else {
 507             return s[ 0 ];
 508         }
 509     }
 510
 511     private static String processSimilarTo( final String name ) {
 512         final int i = name.toLowerCase().indexOf( "similar to" );
 513         String similar_to = "";
 514         if ( i >= 0 ) {
 515             similar_to = " similarity=" + name.substring( i + 10 ).trim();
 516         }
 517         final String pi = processNameIntelligently( name );
 518         return pi + similar_to;
 519     }
 520
 521     private static String sanitize( String s ) {
 522         s = s.replace( ' ', '_' );
 523         s = s.replace( '(', '{' );
 524         s = s.replace( ')', '}' );
 525         s = s.replace( '[', '{' );
 526         s = s.replace( ']', '}' );
 527         s = s.replace( ',', '_' );
 528         return s;
 529     }
 530
 531     public static enum FIELD {
 532         NODE_NAME, SEQUENCE_ANNOTATION_DESC, DOMAIN_STRUCTURE, TAXONOMY_CODE, TAXONOMY_SCIENTIFIC_NAME, SEQUENCE_NAME;
 533     }
 534 }