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: https://sites.google.com/site/cmzmasek/home/software/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
  34 import org.forester.io.parsers.nhx.NHXFormatException;
  35 import org.forester.io.parsers.phyloxml.PhyloXmlDataFormatException;
  36 import org.forester.io.parsers.util.ParserUtils;
  37 import org.forester.phylogeny.Phylogeny;
  38 import org.forester.phylogeny.PhylogenyNode;
  39 import org.forester.phylogeny.data.Accession;
  40 import org.forester.phylogeny.data.Annotation;
  41 import org.forester.phylogeny.data.DomainArchitecture;
  42 import org.forester.phylogeny.data.Identifier;
  43 import org.forester.phylogeny.data.Sequence;
  44 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
  45 import org.forester.util.BasicTable;
  46 import org.forester.util.BasicTableParser;
  47 import org.forester.util.ForesterUtil;
  48
  49 public final class PhylogenyDecorator {
  50
  51     public final static boolean SANITIZE                = false;
  52     final private static String TP_NODE_NAME            = "NODE_NAME";
  53     final private static String TP_SEQ_ACCESSION        = "SEQ_ACCESSION";
  54     final private static String TP_SEQ_ACCESSION_SOURCE = "SEQ_ACCESSION_SOURCE";
  55     final private static String TP_SEQ_ANNOTATION_DESC  = "SEQ_ANNOTATION_DESC";
  56     final private static String TP_SEQ_ANNOTATION_REF   = "SEQ_ANNOTATION_REF";
  57     final private static String TP_SEQ_MOL_SEQ          = "SEQ_MOL_SEQ";
  58     final private static String TP_SEQ_NAME             = "SEQ_NAME";
  59     final private static String TP_SEQ_SYMBOL           = "SEQ_SYMBOL";
  60     final private static String TP_TAXONOMY_CN          = "TAXONOMY_CN";
  61     // From evoruby/lib/evo/apps/tseq_taxonomy_processor.rb:
  62     final private static String TP_TAXONOMY_CODE        = "TAXONOMY_CODE";
  63     final private static String TP_TAXONOMY_ID          = "TAXONOMY_ID";
  64     final private static String TP_TAXONOMY_ID_PROVIDER = "TAXONOMY_ID_PROVIDER";
  65     final private static String TP_TAXONOMY_SN          = "TAXONOMY_SN";
  66     final private static String TP_TAXONOMY_SYN         = "TAXONOMY_SYN";
  67
  68     private PhylogenyDecorator() {
  69         // Not needed.
  70     }
  71
  72     public static void decorate( final Phylogeny phylogeny,
  73                                  final Map<String, Map<String, String>> map,
  74                                  final boolean picky,
  75                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map )
  76             throws IllegalArgumentException, PhyloXmlDataFormatException {
  77         for( final PhylogenyNodeIterator iter = phylogeny.iteratorPostorder(); iter.hasNext(); ) {
  78             final PhylogenyNode node = iter.next();
  79             final String name = node.getName();
  80             if ( !ForesterUtil.isEmpty( name ) ) {
  81                 if ( map.containsKey( name ) || ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 ) ) {
  82                     Map<String, String> new_values = map.get( name );
  83                     int x = 0;
  84                     while ( ( new_values == null ) && ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 )
  85                             && ( x <= numbers_of_chars_allowed_to_remove_if_not_found_in_map ) ) {
  86                         new_values = map.get( name.substring( 0, name.length() - x ) );
  87                         ++x;
  88                     }
  89                     if ( new_values != null ) {
  90                         if ( new_values.containsKey( TP_TAXONOMY_CODE ) ) {
  91                             ForesterUtil.ensurePresenceOfTaxonomy( node );
  92                             node.getNodeData().getTaxonomy().setTaxonomyCode( new_values.get( TP_TAXONOMY_CODE ) );
  93                         }
  94                         if ( new_values.containsKey( TP_TAXONOMY_ID )
  95                                 && new_values.containsKey( TP_TAXONOMY_ID_PROVIDER ) ) {
  96                             ForesterUtil.ensurePresenceOfTaxonomy( node );
  97                             node.getNodeData()
  98                                     .getTaxonomy()
  99                                     .setIdentifier( new Identifier( new_values.get( TP_TAXONOMY_ID ),
 100                                                                     new_values.get( TP_TAXONOMY_ID_PROVIDER ) ) );
 101                         }
 102                         else if ( new_values.containsKey( TP_TAXONOMY_ID ) ) {
 103                             ForesterUtil.ensurePresenceOfTaxonomy( node );
 104                             node.getNodeData().getTaxonomy()
 105                                     .setIdentifier( new Identifier( new_values.get( TP_TAXONOMY_ID ) ) );
 106                         }
 107                         if ( new_values.containsKey( TP_TAXONOMY_SN ) ) {
 108                             ForesterUtil.ensurePresenceOfTaxonomy( node );
 109                             node.getNodeData().getTaxonomy().setScientificName( new_values.get( TP_TAXONOMY_SN ) );
 110                         }
 111                         if ( new_values.containsKey( TP_TAXONOMY_CN ) ) {
 112                             ForesterUtil.ensurePresenceOfTaxonomy( node );
 113                             node.getNodeData().getTaxonomy().setCommonName( new_values.get( TP_TAXONOMY_CN ) );
 114                         }
 115                         if ( new_values.containsKey( TP_TAXONOMY_SYN ) ) {
 116                             ForesterUtil.ensurePresenceOfTaxonomy( node );
 117                             node.getNodeData().getTaxonomy().getSynonyms().add( new_values.get( TP_TAXONOMY_SYN ) );
 118                         }
 119                         if ( new_values.containsKey( TP_SEQ_ACCESSION )
 120                                 && new_values.containsKey( TP_SEQ_ACCESSION_SOURCE ) ) {
 121                             ForesterUtil.ensurePresenceOfSequence( node );
 122                             node.getNodeData()
 123                                     .getSequence()
 124                                     .setAccession( new Accession( new_values.get( TP_SEQ_ACCESSION ),
 125                                                                   new_values.get( TP_SEQ_ACCESSION_SOURCE ) ) );
 126                         }
 127                         if ( new_values.containsKey( TP_SEQ_ANNOTATION_DESC ) ) {
 128                             ForesterUtil.ensurePresenceOfSequence( node );
 129                             final Annotation ann = new Annotation();
 130                             ann.setDesc( new_values.get( TP_SEQ_ANNOTATION_DESC ) );
 131                             node.getNodeData().getSequence().addAnnotation( ann );
 132                         }
 133                         if ( new_values.containsKey( TP_SEQ_ANNOTATION_REF ) ) {
 134                             ForesterUtil.ensurePresenceOfSequence( node );
 135                             final Annotation ann = new Annotation( new_values.get( TP_SEQ_ANNOTATION_REF ) );
 136                             node.getNodeData().getSequence().addAnnotation( ann );
 137                         }
 138                         if ( new_values.containsKey( TP_SEQ_SYMBOL ) ) {
 139                             ForesterUtil.ensurePresenceOfSequence( node );
 140                             node.getNodeData().getSequence().setSymbol( new_values.get( TP_SEQ_SYMBOL ) );
 141                         }
 142                         if ( new_values.containsKey( TP_SEQ_NAME ) ) {
 143                             ForesterUtil.ensurePresenceOfSequence( node );
 144                             node.getNodeData().getSequence().setName( new_values.get( TP_SEQ_NAME ) );
 145                         }
 146                         if ( new_values.containsKey( TP_SEQ_MOL_SEQ ) ) {
 147                             ForesterUtil.ensurePresenceOfSequence( node );
 148                             node.getNodeData().getSequence().setMolecularSequence( new_values.get( TP_SEQ_MOL_SEQ ) );
 149                         }
 150                         if ( new_values.containsKey( TP_NODE_NAME ) ) {
 151                             node.setName( new_values.get( TP_NODE_NAME ) );
 152                         }
 153                     } // if ( new_values != null )
 154                 } // if ( map.containsKey( name ) || ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 ) )
 155                 else if ( picky ) {
 156                     throw new IllegalArgumentException( "\"" + name + "\" not found in name map" );
 157                 }
 158             }
 159         }
 160     }
 161
 162     public static void decorate( final Phylogeny phylogeny,
 163                                  final Map<String, String> map,
 164                                  final FIELD field,
 165                                  final boolean extract_bracketed_scientific_name,
 166                                  final boolean extract_bracketed_tax_code,
 167                                  final boolean picky,
 168                                  final boolean cut_name_after_space,
 169                                  final boolean process_name_intelligently,
 170                                  final boolean process_similar_to,
 171                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 172                                  final boolean trim_after_tilde,
 173                                  final boolean verbose ) throws IllegalArgumentException, NHXFormatException,
 174             PhyloXmlDataFormatException {
 175         PhylogenyDecorator.decorate( phylogeny,
 176                                      map,
 177                                      field,
 178                                      extract_bracketed_scientific_name,
 179                                      extract_bracketed_tax_code,
 180                                      picky,
 181                                      null,
 182                                      cut_name_after_space,
 183                                      process_name_intelligently,
 184                                      process_similar_to,
 185                                      numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 186                                      trim_after_tilde,
 187                                      verbose );
 188     }
 189
 190     /**
 191      *
 192      *
 193      *
 194      * @param phylogeny
 195      * @param map
 196      *            maps names (in phylogeny) to new values if intermediate_map is
 197      *            null otherwise maps intermediate value to new value
 198      * @param field
 199      * @param picky
 200      * @param intermediate_map
 201      *            maps name (in phylogeny) to a intermediate value
 202      * @throws IllegalArgumentException
 203      * @throws PhyloXmlDataFormatException
 204      */
 205     public static void decorate( final Phylogeny phylogeny,
 206                                  final Map<String, String> map,
 207                                  final FIELD field,
 208                                  final boolean extract_bracketed_scientific_name,
 209                                  final boolean extract_bracketed_tax_code,
 210                                  final boolean picky,
 211                                  final Map<String, String> intermediate_map,
 212                                  final boolean cut_name_after_space,
 213                                  final boolean process_name_intelligently,
 214                                  final boolean process_similar_to,
 215                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 216                                  final boolean trim_after_tilde,
 217                                  final boolean verbose ) throws IllegalArgumentException, PhyloXmlDataFormatException {
 218         if ( extract_bracketed_scientific_name && ( field == FIELD.TAXONOMY_SCIENTIFIC_NAME ) ) {
 219             throw new IllegalArgumentException( "attempt to extract bracketed scientific name together with data field pointing to scientific name" );
 220         }
 221         if ( map.isEmpty() ) {
 222             throw new IllegalArgumentException( "map is empty" );
 223         }
 224         for( final PhylogenyNodeIterator iter = phylogeny.iteratorPostorder(); iter.hasNext(); ) {
 225             final PhylogenyNode node = iter.next();
 226             String name = node.getName();
 227             String tilde_annotation = null;
 228             if ( trim_after_tilde && ( name.indexOf( '~' ) > 0 ) ) {
 229                 final int ti = name.indexOf( '~' );
 230                 tilde_annotation = name.substring( ti );
 231                 name = name.substring( 0, ti );
 232             }
 233             if ( !ForesterUtil.isEmpty( name ) ) {
 234                 if ( intermediate_map != null ) {
 235                     name = PhylogenyDecorator.extractIntermediate( intermediate_map, name, verbose );
 236                 }
 237                 if ( map.containsKey( name ) || ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 ) ) {
 238                     String new_value = map.get( name );
 239                     int x = 0;
 240                     while ( ( new_value == null ) && ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 )
 241                             && ( x <= numbers_of_chars_allowed_to_remove_if_not_found_in_map ) ) {
 242                         new_value = map.get( name.substring( 0, name.length() - x ) );
 243                         ++x;
 244                     }
 245                     if ( new_value != null ) {
 246                         new_value = new_value.trim();
 247                         new_value.replaceAll( "/\\s+/", " " );
 248                         if ( extract_bracketed_scientific_name && new_value.endsWith( "]" ) ) {
 249                             new_value = extractBracketedScientificNames( node, new_value );
 250                         }
 251                         else if ( extract_bracketed_tax_code ) {
 252                             if ( ParserUtils.TAXOMONY_CODE_PATTERN_BRACKETED.matcher( new_value ).find() ) {
 253                                 new_value = extractBracketedTaxCodes( node, new_value );
 254                             }
 255                             else if ( picky ) {
 256                                 throw new IllegalArgumentException( " could not get taxonomy from \"" + new_value
 257                                         + "\"" );
 258                             }
 259                         }
 260                         switch ( field ) {
 261                             case MOL_SEQ:
 262                                 if ( verbose ) {
 263                                     System.out.println( name + ": " + new_value );
 264                                 }
 265                                 if ( !node.getNodeData().isHasSequence() ) {
 266                                     node.getNodeData().setSequence( new Sequence() );
 267                                 }
 268                                 node.getNodeData().getSequence().setMolecularSequence( new_value );
 269                                 break;
 270                             case SEQUENCE_ANNOTATION_DESC:
 271                                 if ( verbose ) {
 272                                     System.out.println( name + ": " + new_value );
 273                                 }
 274                                 if ( !node.getNodeData().isHasSequence() ) {
 275                                     node.getNodeData().setSequence( new Sequence() );
 276                                 }
 277                                 final Annotation annotation = new Annotation();
 278                                 annotation.setDesc( new_value );
 279                                 node.getNodeData().getSequence().addAnnotation( annotation );
 280                                 break;
 281                             case DOMAIN_STRUCTURE:
 282                                 if ( verbose ) {
 283                                     System.out.println( name + ": " + new_value );
 284                                 }
 285                                 if ( !node.getNodeData().isHasSequence() ) {
 286                                     node.getNodeData().setSequence( new Sequence() );
 287                                 }
 288                                 node.getNodeData().getSequence()
 289                                         .setDomainArchitecture( new DomainArchitecture( new_value ) );
 290                                 break;
 291                             case TAXONOMY_CODE:
 292                                 if ( verbose ) {
 293                                     System.out.println( name + ": " + new_value );
 294                                 }
 295                                 ForesterUtil.ensurePresenceOfTaxonomy( node );
 296                                 node.getNodeData().getTaxonomy().setTaxonomyCode( new_value );
 297                                 break;
 298                             case TAXONOMY_SCIENTIFIC_NAME:
 299                                 if ( verbose ) {
 300                                     System.out.println( name + ": " + new_value );
 301                                 }
 302                                 ForesterUtil.ensurePresenceOfTaxonomy( node );
 303                                 node.getNodeData().getTaxonomy().setScientificName( new_value );
 304                                 break;
 305                             case SEQUENCE_NAME:
 306                                 if ( trim_after_tilde ) {
 307                                     new_value = addTildeAnnotation( tilde_annotation, new_value );
 308                                 }
 309                                 if ( verbose ) {
 310                                     System.out.println( name + ": " + new_value );
 311                                 }
 312                                 if ( !node.getNodeData().isHasSequence() ) {
 313                                     node.getNodeData().setSequence( new Sequence() );
 314                                 }
 315                                 node.getNodeData().getSequence().setName( new_value );
 316                                 break;
 317                             case NODE_NAME:
 318                                 if ( verbose ) {
 319                                     System.out.print( name + " -> " );
 320                                 }
 321                                 if ( cut_name_after_space ) {
 322                                     if ( verbose ) {
 323                                         System.out.print( new_value + " -> " );
 324                                     }
 325                                     new_value = PhylogenyDecorator.deleteAtFirstSpace( new_value );
 326                                 }
 327                                 else if ( process_name_intelligently ) {
 328                                     if ( verbose ) {
 329                                         System.out.print( new_value + " -> " );
 330                                     }
 331                                     new_value = PhylogenyDecorator.processNameIntelligently( new_value );
 332                                 }
 333                                 else if ( process_similar_to ) {
 334                                     if ( verbose ) {
 335                                         System.out.print( new_value + " -> " );
 336                                     }
 337                                     new_value = PhylogenyDecorator.processSimilarTo( new_value );
 338                                 }
 339                                 if ( PhylogenyDecorator.SANITIZE ) {
 340                                     new_value = PhylogenyDecorator.sanitize( new_value );
 341                                 }
 342                                 if ( trim_after_tilde ) {
 343                                     new_value = addTildeAnnotation( tilde_annotation, new_value );
 344                                 }
 345                                 if ( verbose ) {
 346                                     System.out.println( new_value );
 347                                 }
 348                                 node.setName( new_value );
 349                                 break;
 350                             default:
 351                                 throw new RuntimeException( "unknown field \"" + field + "\"" );
 352                         }
 353                     }
 354                 }
 355                 else if ( picky ) {
 356                     throw new IllegalArgumentException( "\"" + name + "\" not found in name map" );
 357                 }
 358             }
 359         }
 360     }
 361
 362     public static void decorate( final Phylogeny[] phylogenies,
 363                                  final Map<String, Map<String, String>> map,
 364                                  final boolean picky,
 365                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map )
 366             throws IllegalArgumentException, NHXFormatException, PhyloXmlDataFormatException {
 367         for( final Phylogeny phylogenie : phylogenies ) {
 368             PhylogenyDecorator
 369                     .decorate( phylogenie, map, picky, numbers_of_chars_allowed_to_remove_if_not_found_in_map );
 370         }
 371     }
 372
 373     public static void decorate( final Phylogeny[] phylogenies,
 374                                  final Map<String, String> map,
 375                                  final FIELD field,
 376                                  final boolean extract_bracketed_scientific_name,
 377                                  final boolean extract_bracketed_tax_code,
 378                                  final boolean picky,
 379                                  final boolean cut_name_after_space,
 380                                  final boolean process_name_intelligently,
 381                                  final boolean process_similar_to,
 382                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 383                                  final boolean trim_after_tilde,
 384                                  final boolean verbose ) throws IllegalArgumentException, NHXFormatException,
 385             PhyloXmlDataFormatException {
 386         for( final Phylogeny phylogenie : phylogenies ) {
 387             PhylogenyDecorator.decorate( phylogenie,
 388                                          map,
 389                                          field,
 390                                          extract_bracketed_scientific_name,
 391                                          extract_bracketed_tax_code,
 392                                          picky,
 393                                          cut_name_after_space,
 394                                          process_name_intelligently,
 395                                          process_similar_to,
 396                                          numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 397                                          trim_after_tilde,
 398                                          verbose );
 399         }
 400     }
 401
 402     public static void decorate( final Phylogeny[] phylogenies,
 403                                  final Map<String, String> map,
 404                                  final FIELD field,
 405                                  final boolean extract_bracketed_scientific_name,
 406                                  final boolean extract_bracketed_tax_code,
 407                                  final boolean picky,
 408                                  final Map<String, String> intermediate_map,
 409                                  final boolean cut_name_after_space,
 410                                  final boolean process_name_intelligently,
 411                                  final boolean process_similar_to,
 412                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 413                                  final boolean trim_after_tilde,
 414                                  final boolean verbose ) throws IllegalArgumentException, NHXFormatException,
 415             PhyloXmlDataFormatException {
 416         for( final Phylogeny phylogenie : phylogenies ) {
 417             PhylogenyDecorator.decorate( phylogenie,
 418                                          map,
 419                                          field,
 420                                          extract_bracketed_scientific_name,
 421                                          extract_bracketed_tax_code,
 422                                          picky,
 423                                          intermediate_map,
 424                                          cut_name_after_space,
 425                                          process_name_intelligently,
 426                                          process_similar_to,
 427                                          numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 428                                          trim_after_tilde,
 429                                          verbose );
 430         }
 431     }
 432
 433     public static Map<String, Map<String, String>> parseMappingTable( final File mapping_table_file )
 434             throws IOException {
 435         final Map<String, Map<String, String>> map = new HashMap<String, Map<String, String>>();
 436         BasicTable<String> mapping_table = null;
 437         mapping_table = BasicTableParser.parse( mapping_table_file, '\t', false, false );
 438         for( int row = 0; row < mapping_table.getNumberOfRows(); ++row ) {
 439             final Map<String, String> row_map = new HashMap<String, String>();
 440             String name = null;
 441             for( int col = 0; col < mapping_table.getNumberOfColumns(); ++col ) {
 442                 final String table_cell = mapping_table.getValue( col, row );
 443                 if ( col == 0 ) {
 444                     name = table_cell;
 445                 }
 446                 else if ( table_cell != null ) {
 447                     final String key = table_cell.substring( 0, table_cell.indexOf( ':' ) );
 448                     final String val = table_cell.substring( table_cell.indexOf( ':' ) + 1, table_cell.length() );
 449                     row_map.put( key, val );
 450                 }
 451             }
 452             map.put( name, row_map );
 453         }
 454         return map;
 455     }
 456
 457     private final static String addTildeAnnotation( final String tilde_annotation, final String new_value ) {
 458         if ( ForesterUtil.isEmpty( tilde_annotation ) ) {
 459             return new_value;
 460         }
 461         return new_value + tilde_annotation;
 462     }
 463
 464     private static String deleteAtFirstSpace( final String name ) {
 465         final int first_space = name.indexOf( " " );
 466         if ( first_space > 1 ) {
 467             return name.substring( 0, first_space ).trim();
 468         }
 469         return name;
 470     }
 471
 472     private static String extractBracketedScientificNames( final PhylogenyNode node, final String new_value ) {
 473         final int i = new_value.lastIndexOf( "[" );
 474         final String scientific_name = new_value.substring( i + 1, new_value.length() - 1 );
 475         ForesterUtil.ensurePresenceOfTaxonomy( node );
 476         node.getNodeData().getTaxonomy().setScientificName( scientific_name );
 477         return new_value.substring( 0, i - 1 ).trim();
 478     }
 479
 480     private static String extractBracketedTaxCodes( final PhylogenyNode node, final String new_value ) {
 481         final StringBuilder sb = new StringBuilder();
 482         sb.append( new_value );
 483         final String tc = extractBracketedTaxCodes( sb );
 484         if ( !ForesterUtil.isEmpty( tc ) ) {
 485             ForesterUtil.ensurePresenceOfTaxonomy( node );
 486             try {
 487                 node.getNodeData().getTaxonomy().setTaxonomyCode( tc );
 488             }
 489             catch ( final PhyloXmlDataFormatException e ) {
 490                 throw new IllegalArgumentException( "illegal format for taxonomy code: " + tc );
 491             }
 492             return sb.toString().trim();
 493         }
 494         return new_value;
 495     }
 496
 497     private static String extractBracketedTaxCodes( final StringBuilder sb ) {
 498         final Matcher m = ParserUtils.TAXOMONY_CODE_PATTERN_BRACKETED.matcher( sb );
 499         if ( m.find() ) {
 500             final String tc = m.group( 1 );
 501             sb.delete( m.start( 1 ) - 1, m.end( 1 ) + 1 );
 502             return tc;
 503         }
 504         return null;
 505     }
 506
 507     private static String extractIntermediate( final Map<String, String> intermediate_map,
 508                                                final String name,
 509                                                final boolean verbose ) {
 510         String new_name = null;
 511         if ( verbose ) {
 512             System.out.print( name + " => " );
 513         }
 514         if ( intermediate_map.containsKey( name ) ) {
 515             new_name = intermediate_map.get( name );
 516             if ( ForesterUtil.isEmpty( new_name ) ) {
 517                 throw new IllegalArgumentException( "\"" + name + "\" maps to null or empty string in secondary map" );
 518             }
 519         }
 520         else {
 521             throw new IllegalArgumentException( "\"" + name + "\" not found in name secondary map" );
 522         }
 523         if ( verbose ) {
 524             System.out.println( new_name + "  " );
 525         }
 526         return new_name;
 527     }
 528
 529     private static String processNameIntelligently( final String name ) {
 530         final String[] s = name.split( " " );
 531         if ( s.length < 2 ) {
 532             return name;
 533         }
 534         else if ( ( s[ 0 ].indexOf( "_" ) > 0 ) && ( s[ 0 ].indexOf( "|" ) > 0 ) ) {
 535             return s[ 0 ];
 536         }
 537         else if ( ( s[ 1 ].indexOf( "_" ) > 0 ) && ( s[ 1 ].indexOf( "|" ) > 0 ) ) {
 538             return s[ 1 ];
 539         }
 540         else if ( ( s[ 0 ].indexOf( "_" ) > 0 ) && ( s[ 0 ].indexOf( "." ) > 0 ) ) {
 541             return s[ 0 ];
 542         }
 543         else if ( ( s[ 1 ].indexOf( "_" ) > 0 ) && ( s[ 1 ].indexOf( "." ) > 0 ) ) {
 544             return s[ 1 ];
 545         }
 546         else if ( s[ 0 ].indexOf( "_" ) > 0 ) {
 547             return s[ 0 ];
 548         }
 549         else if ( s[ 1 ].indexOf( "_" ) > 0 ) {
 550             return s[ 1 ];
 551         }
 552         else {
 553             return s[ 0 ];
 554         }
 555     }
 556
 557     private static String processSimilarTo( final String name ) {
 558         final int i = name.toLowerCase().indexOf( "similar to" );
 559         String similar_to = "";
 560         if ( i >= 0 ) {
 561             similar_to = " similarity=" + name.substring( i + 10 ).trim();
 562         }
 563         final String pi = processNameIntelligently( name );
 564         return pi + similar_to;
 565     }
 566
 567     private static String sanitize( String s ) {
 568         s = s.replace( ' ', '_' );
 569         s = s.replace( '(', '{' );
 570         s = s.replace( ')', '}' );
 571         s = s.replace( '[', '{' );
 572         s = s.replace( ']', '}' );
 573         s = s.replace( ',', '_' );
 574         return s;
 575     }
 576
 577     public static enum FIELD {
 578         DOMAIN_STRUCTURE,
 579         MOL_SEQ,
 580         NODE_NAME,
 581         SEQUENCE_ANNOTATION_DESC,
 582         SEQUENCE_NAME,
 583         TAXONOMY_CODE,
 584         TAXONOMY_SCIENTIFIC_NAME;
 585     }
 586 }