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     // From evoruby/lib/evo/apps/tseq_taxonomy_processor.rb:
  52     final private static String TP_TAXONOMY_CODE        = "TAXONOMY_CODE";
  53     final private static String TP_TAXONOMY_ID          = "TAXONOMY_ID";
  54     final private static String TP_TAXONOMY_ID_PROVIDER = "TAXONOMY_ID_PROVIDER";
  55     final private static String TP_TAXONOMY_SN          = "TAXONOMY_SN";
  56     final private static String TP_TAXONOMY_CN          = "TAXONOMY_CN";
  57     final private static String TP_TAXONOMY_SYN         = "TAXONOMY_SYN";
  58     final private static String TP_SEQ_SYMBOL           = "SEQ_SYMBOL";
  59     final private static String TP_SEQ_ACCESSION        = "SEQ_ACCESSION";
  60     final private static String TP_SEQ_ACCESSION_SOURCE = "SEQ_ACCESSION_SOURCE";
  61     final private static String TP_SEQ_ANNOTATION_DESC  = "SEQ_ANNOTATION_DESC";
  62     final private static String TP_SEQ_ANNOTATION_REF   = "SEQ_ANNOTATION_REF";
  63     final private static String TP_SEQ_MOL_SEQ          = "SEQ_MOL_SEQ";
  64     final private static String TP_SEQ_NAME             = "SEQ_NAME";
  65     final private static String TP_NODE_NAME            = "NODE_NAME";
  66     public final static boolean SANITIZE                = false;
  67     public final static boolean VERBOSE                 = true;
  68
  69     private PhylogenyDecorator() {
  70         // Not needed.
  71     }
  72
  73     public static void decorate( final Phylogeny phylogeny,
  74                                  final Map<String, Map<String, String>> map,
  75                                  final boolean picky,
  76                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map )
  77             throws IllegalArgumentException, PhyloXmlDataFormatException {
  78         for( final PhylogenyNodeIterator iter = phylogeny.iteratorPostorder(); iter.hasNext(); ) {
  79             final PhylogenyNode node = iter.next();
  80             final String name = node.getName();
  81             if ( !ForesterUtil.isEmpty( name ) ) {
  82                 if ( map.containsKey( name ) || ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 ) ) {
  83                     Map<String, String> new_values = map.get( name );
  84                     int x = 0;
  85                     while ( ( new_values == null ) && ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 )
  86                             && ( x <= numbers_of_chars_allowed_to_remove_if_not_found_in_map ) ) {
  87                         new_values = map.get( name.substring( 0, name.length() - x ) );
  88                         ++x;
  89                     }
  90                     if ( new_values != null ) {
  91                         if ( new_values.containsKey( TP_TAXONOMY_CODE ) ) {
  92                             ForesterUtil.ensurePresenceOfTaxonomy( node );
  93                             node.getNodeData().getTaxonomy().setTaxonomyCode( new_values.get( TP_TAXONOMY_CODE ) );
  94                         }
  95                         if ( new_values.containsKey( TP_TAXONOMY_ID )
  96                                 && new_values.containsKey( TP_TAXONOMY_ID_PROVIDER ) ) {
  97                             ForesterUtil.ensurePresenceOfTaxonomy( node );
  98                             node.getNodeData()
  99                                     .getTaxonomy()
 100                                     .setIdentifier( new Identifier( new_values.get( TP_TAXONOMY_ID ),
 101                                                                     new_values.get( TP_TAXONOMY_ID_PROVIDER ) ) );
 102                         }
 103                         else if ( new_values.containsKey( TP_TAXONOMY_ID ) ) {
 104                             ForesterUtil.ensurePresenceOfTaxonomy( node );
 105                             node.getNodeData().getTaxonomy()
 106                                     .setIdentifier( new Identifier( new_values.get( TP_TAXONOMY_ID ) ) );
 107                         }
 108                         if ( new_values.containsKey( TP_TAXONOMY_SN ) ) {
 109                             ForesterUtil.ensurePresenceOfTaxonomy( node );
 110                             node.getNodeData().getTaxonomy().setScientificName( new_values.get( TP_TAXONOMY_SN ) );
 111                         }
 112                         if ( new_values.containsKey( TP_TAXONOMY_CN ) ) {
 113                             ForesterUtil.ensurePresenceOfTaxonomy( node );
 114                             node.getNodeData().getTaxonomy().setCommonName( new_values.get( TP_TAXONOMY_CN ) );
 115                         }
 116                         if ( new_values.containsKey( TP_TAXONOMY_SYN ) ) {
 117                             ForesterUtil.ensurePresenceOfTaxonomy( node );
 118                             node.getNodeData().getTaxonomy().getSynonyms().add( new_values.get( TP_TAXONOMY_SYN ) );
 119                         }
 120                         if ( new_values.containsKey( TP_SEQ_ACCESSION )
 121                                 && new_values.containsKey( TP_SEQ_ACCESSION_SOURCE ) ) {
 122                             ForesterUtil.ensurePresenceOfSequence( node );
 123                             node.getNodeData()
 124                                     .getSequence()
 125                                     .setAccession( new Accession( new_values.get( TP_SEQ_ACCESSION ),
 126                                                                   new_values.get( TP_SEQ_ACCESSION_SOURCE ) ) );
 127                         }
 128                         if ( new_values.containsKey( TP_SEQ_ANNOTATION_DESC ) ) {
 129                             ForesterUtil.ensurePresenceOfSequence( node );
 130                             final Annotation ann = new Annotation();
 131                             ann.setDesc( new_values.get( TP_SEQ_ANNOTATION_DESC ) );
 132                             node.getNodeData().getSequence().addAnnotation( ann );
 133                         }
 134                         if ( new_values.containsKey( TP_SEQ_ANNOTATION_REF ) ) {
 135                             ForesterUtil.ensurePresenceOfSequence( node );
 136                             final Annotation ann = new Annotation( new_values.get( TP_SEQ_ANNOTATION_REF ) );
 137                             node.getNodeData().getSequence().addAnnotation( ann );
 138                         }
 139                         if ( new_values.containsKey( TP_SEQ_SYMBOL ) ) {
 140                             ForesterUtil.ensurePresenceOfSequence( node );
 141                             node.getNodeData().getSequence().setSymbol( new_values.get( TP_SEQ_SYMBOL ) );
 142                         }
 143                         if ( new_values.containsKey( TP_SEQ_NAME ) ) {
 144                             ForesterUtil.ensurePresenceOfSequence( node );
 145                             node.getNodeData().getSequence().setName( new_values.get( TP_SEQ_NAME ) );
 146                         }
 147                         if ( new_values.containsKey( TP_SEQ_MOL_SEQ ) ) {
 148                             ForesterUtil.ensurePresenceOfSequence( node );
 149                             node.getNodeData().getSequence().setMolecularSequence( new_values.get( TP_SEQ_MOL_SEQ ) );
 150                         }
 151                         if ( new_values.containsKey( TP_NODE_NAME ) ) {
 152                             node.setName( new_values.get( TP_NODE_NAME ) );
 153                         }
 154                     } // if ( new_values != null )
 155                 } // if ( map.containsKey( name ) || ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 ) )
 156                 else if ( picky ) {
 157                     throw new IllegalArgumentException( "\"" + name + "\" not found in name map" );
 158                 }
 159             }
 160         }
 161     }
 162
 163     public static void decorate( final Phylogeny phylogeny,
 164                                  final Map<String, String> map,
 165                                  final FIELD field,
 166                                  final boolean extract_bracketed_scientific_name,
 167                                  final boolean extract_bracketed_tax_code,
 168                                  final boolean picky,
 169                                  final boolean cut_name_after_space,
 170                                  final boolean process_name_intelligently,
 171                                  final boolean process_similar_to,
 172                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 173                                  final boolean trim_after_tilde ) 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     }
 188
 189     /**
 190      *
 191      *
 192      *
 193      * @param phylogeny
 194      * @param map
 195      *            maps names (in phylogeny) to new values if intermediate_map is
 196      *            null otherwise maps intermediate value to new value
 197      * @param field
 198      * @param picky
 199      * @param intermediate_map
 200      *            maps name (in phylogeny) to a intermediate value
 201      * @throws IllegalArgumentException
 202      * @throws PhyloXmlDataFormatException
 203      */
 204     public static void decorate( final Phylogeny phylogeny,
 205                                  final Map<String, String> map,
 206                                  final FIELD field,
 207                                  final boolean extract_bracketed_scientific_name,
 208                                  final boolean extract_bracketed_tax_code,
 209                                  final boolean picky,
 210                                  final Map<String, String> intermediate_map,
 211                                  final boolean cut_name_after_space,
 212                                  final boolean process_name_intelligently,
 213                                  final boolean process_similar_to,
 214                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 215                                  final boolean trim_after_tilde ) throws IllegalArgumentException,
 216             PhyloXmlDataFormatException {
 217         if ( extract_bracketed_scientific_name && ( field == FIELD.TAXONOMY_SCIENTIFIC_NAME ) ) {
 218             throw new IllegalArgumentException( "attempt to extract bracketed scientific name together with data field pointing to scientific name" );
 219         }
 220         for( final PhylogenyNodeIterator iter = phylogeny.iteratorPostorder(); iter.hasNext(); ) {
 221             final PhylogenyNode node = iter.next();
 222             String name = node.getName();
 223             String tilde_annotation = null;
 224             if ( trim_after_tilde && ( name.indexOf( '~' ) > 0 ) ) {
 225                 final int ti = name.indexOf( '~' );
 226                 tilde_annotation = name.substring( ti );
 227                 name = name.substring( 0, ti );
 228             }
 229             if ( !ForesterUtil.isEmpty( name ) ) {
 230                 if ( intermediate_map != null ) {
 231                     name = PhylogenyDecorator.extractIntermediate( intermediate_map, name );
 232                 }
 233                 if ( map.containsKey( name ) || ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 ) ) {
 234                     String new_value = map.get( name );
 235                     int x = 0;
 236                     while ( ( new_value == null ) && ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 )
 237                             && ( x <= numbers_of_chars_allowed_to_remove_if_not_found_in_map ) ) {
 238                         new_value = map.get( name.substring( 0, name.length() - x ) );
 239                         ++x;
 240                     }
 241                     if ( new_value != null ) {
 242                         new_value = new_value.trim();
 243                         new_value.replaceAll( "/\\s+/", " " );
 244                         if ( extract_bracketed_scientific_name && new_value.endsWith( "]" ) ) {
 245                             new_value = extractBracketedScientificNames( node, new_value );
 246                         }
 247                         else if ( extract_bracketed_tax_code
 248                                 && ParserUtils.TAXOMONY_CODE_PATTERN_4.matcher( new_value ).matches() ) {
 249                             new_value = extractBracketedTaxCodes( node, new_value );
 250                         }
 251                         switch ( field ) {
 252                             case SEQUENCE_ANNOTATION_DESC:
 253                                 if ( PhylogenyDecorator.VERBOSE ) {
 254                                     System.out.println( name + ": " + new_value );
 255                                 }
 256                                 if ( !node.getNodeData().isHasSequence() ) {
 257                                     node.getNodeData().setSequence( new Sequence() );
 258                                 }
 259                                 final Annotation annotation = new Annotation( "?" );
 260                                 annotation.setDesc( new_value );
 261                                 node.getNodeData().getSequence().addAnnotation( annotation );
 262                                 break;
 263                             case DOMAIN_STRUCTURE:
 264                                 if ( PhylogenyDecorator.VERBOSE ) {
 265                                     System.out.println( name + ": " + new_value );
 266                                 }
 267                                 if ( !node.getNodeData().isHasSequence() ) {
 268                                     node.getNodeData().setSequence( new Sequence() );
 269                                 }
 270                                 node.getNodeData().getSequence()
 271                                         .setDomainArchitecture( new DomainArchitecture( new_value ) );
 272                                 break;
 273                             case TAXONOMY_CODE:
 274                                 if ( PhylogenyDecorator.VERBOSE ) {
 275                                     System.out.println( name + ": " + new_value );
 276                                 }
 277                                 ForesterUtil.ensurePresenceOfTaxonomy( node );
 278                                 node.getNodeData().getTaxonomy().setTaxonomyCode( new_value );
 279                                 break;
 280                             case TAXONOMY_SCIENTIFIC_NAME:
 281                                 if ( PhylogenyDecorator.VERBOSE ) {
 282                                     System.out.println( name + ": " + new_value );
 283                                 }
 284                                 ForesterUtil.ensurePresenceOfTaxonomy( node );
 285                                 node.getNodeData().getTaxonomy().setScientificName( new_value );
 286                                 break;
 287                             case SEQUENCE_NAME:
 288                                 if ( trim_after_tilde ) {
 289                                     new_value = addTildeAnnotation( tilde_annotation, new_value );
 290                                 }
 291                                 if ( PhylogenyDecorator.VERBOSE ) {
 292                                     System.out.println( name + ": " + new_value );
 293                                 }
 294                                 if ( !node.getNodeData().isHasSequence() ) {
 295                                     node.getNodeData().setSequence( new Sequence() );
 296                                 }
 297                                 node.getNodeData().getSequence().setName( new_value );
 298                                 break;
 299                             case NODE_NAME:
 300                                 if ( PhylogenyDecorator.VERBOSE ) {
 301                                     System.out.print( name + " -> " );
 302                                 }
 303                                 if ( cut_name_after_space ) {
 304                                     if ( PhylogenyDecorator.VERBOSE ) {
 305                                         System.out.print( new_value + " -> " );
 306                                     }
 307                                     new_value = PhylogenyDecorator.deleteAtFirstSpace( new_value );
 308                                 }
 309                                 else if ( process_name_intelligently ) {
 310                                     if ( PhylogenyDecorator.VERBOSE ) {
 311                                         System.out.print( new_value + " -> " );
 312                                     }
 313                                     new_value = PhylogenyDecorator.processNameIntelligently( new_value );
 314                                 }
 315                                 else if ( process_similar_to ) {
 316                                     if ( PhylogenyDecorator.VERBOSE ) {
 317                                         System.out.print( new_value + " -> " );
 318                                     }
 319                                     new_value = PhylogenyDecorator.processSimilarTo( new_value );
 320                                 }
 321                                 if ( PhylogenyDecorator.SANITIZE ) {
 322                                     new_value = PhylogenyDecorator.sanitize( new_value );
 323                                 }
 324                                 if ( trim_after_tilde ) {
 325                                     new_value = addTildeAnnotation( tilde_annotation, new_value );
 326                                 }
 327                                 if ( PhylogenyDecorator.VERBOSE ) {
 328                                     System.out.println( new_value );
 329                                 }
 330                                 node.setName( new_value );
 331                                 break;
 332                             default:
 333                                 throw new RuntimeException( "unknown field \"" + field + "\"" );
 334                         }
 335                     }
 336                 }
 337                 else if ( picky ) {
 338                     throw new IllegalArgumentException( "\"" + name + "\" not found in name map" );
 339                 }
 340             }
 341         }
 342     }
 343
 344     private final static String addTildeAnnotation( final String tilde_annotation, final String new_value ) {
 345         if ( ForesterUtil.isEmpty( tilde_annotation ) ) {
 346             return new_value;
 347         }
 348         return new_value + tilde_annotation;
 349     }
 350
 351     public static void decorate( final Phylogeny[] phylogenies,
 352                                  final Map<String, Map<String, String>> map,
 353                                  final boolean picky,
 354                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map )
 355             throws IllegalArgumentException, NHXFormatException, PhyloXmlDataFormatException {
 356         for( final Phylogeny phylogenie : phylogenies ) {
 357             PhylogenyDecorator
 358                     .decorate( phylogenie, map, picky, numbers_of_chars_allowed_to_remove_if_not_found_in_map );
 359         }
 360     }
 361
 362     public static void decorate( final Phylogeny[] phylogenies,
 363                                  final Map<String, String> map,
 364                                  final FIELD field,
 365                                  final boolean extract_bracketed_scientific_name,
 366                                  final boolean extract_bracketed_tax_code,
 367                                  final boolean picky,
 368                                  final boolean cut_name_after_space,
 369                                  final boolean process_name_intelligently,
 370                                  final boolean process_similar_to,
 371                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 372                                  final boolean trim_after_tilde ) throws IllegalArgumentException, NHXFormatException,
 373             PhyloXmlDataFormatException {
 374         for( final Phylogeny phylogenie : phylogenies ) {
 375             PhylogenyDecorator.decorate( phylogenie,
 376                                          map,
 377                                          field,
 378                                          extract_bracketed_scientific_name,
 379                                          extract_bracketed_tax_code,
 380                                          picky,
 381                                          cut_name_after_space,
 382                                          process_name_intelligently,
 383                                          process_similar_to,
 384                                          numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 385                                          trim_after_tilde );
 386         }
 387     }
 388
 389     public static void decorate( final Phylogeny[] phylogenies,
 390                                  final Map<String, String> map,
 391                                  final FIELD field,
 392                                  final boolean extract_bracketed_scientific_name,
 393                                  final boolean extract_bracketed_tax_code,
 394                                  final boolean picky,
 395                                  final Map<String, String> intermediate_map,
 396                                  final boolean cut_name_after_space,
 397                                  final boolean process_name_intelligently,
 398                                  final boolean process_similar_to,
 399                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 400                                  final boolean trim_after_tilde ) throws IllegalArgumentException, NHXFormatException,
 401             PhyloXmlDataFormatException {
 402         for( final Phylogeny phylogenie : phylogenies ) {
 403             PhylogenyDecorator.decorate( phylogenie,
 404                                          map,
 405                                          field,
 406                                          extract_bracketed_scientific_name,
 407                                          extract_bracketed_tax_code,
 408                                          picky,
 409                                          intermediate_map,
 410                                          cut_name_after_space,
 411                                          process_name_intelligently,
 412                                          process_similar_to,
 413                                          numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 414                                          trim_after_tilde );
 415         }
 416     }
 417
 418     public static Map<String, Map<String, String>> parseMappingTable( final File mapping_table_file )
 419             throws IOException {
 420         final Map<String, Map<String, String>> map = new HashMap<String, Map<String, String>>();
 421         BasicTable<String> mapping_table = null;
 422         mapping_table = BasicTableParser.parse( mapping_table_file, "\t", false, false );
 423         for( int row = 0; row < mapping_table.getNumberOfRows(); ++row ) {
 424             final Map<String, String> row_map = new HashMap<String, String>();
 425             String name = null;
 426             for( int col = 0; col < mapping_table.getNumberOfColumns(); ++col ) {
 427                 final String table_cell = mapping_table.getValue( col, row );
 428                 if ( col == 0 ) {
 429                     name = table_cell;
 430                 }
 431                 else if ( table_cell != null ) {
 432                     final String key = table_cell.substring( 0, table_cell.indexOf( ':' ) );
 433                     final String val = table_cell.substring( table_cell.indexOf( ':' ) + 1, table_cell.length() );
 434                     row_map.put( key, val );
 435                 }
 436             }
 437             map.put( name, row_map );
 438         }
 439         return map;
 440     }
 441
 442     private static String deleteAtFirstSpace( final String name ) {
 443         final int first_space = name.indexOf( " " );
 444         if ( first_space > 1 ) {
 445             return name.substring( 0, first_space ).trim();
 446         }
 447         return name;
 448     }
 449
 450     private static String extractBracketedScientificNames( final PhylogenyNode node, final String new_value ) {
 451         final int i = new_value.lastIndexOf( "[" );
 452         final String scientific_name = new_value.substring( i + 1, new_value.length() - 1 );
 453         ForesterUtil.ensurePresenceOfTaxonomy( node );
 454         node.getNodeData().getTaxonomy().setScientificName( scientific_name );
 455         return new_value.substring( 0, i - 1 ).trim();
 456     }
 457
 458     private static String extractBracketedTaxCodes( final PhylogenyNode node, final String new_value ) {
 459         final Matcher m = ParserUtils.TAXOMONY_CODE_PATTERN_4.matcher( new_value );
 460         String tc = null;
 461         if ( m.matches() ) {
 462             tc = m.group( 1 );
 463         }
 464         ForesterUtil.ensurePresenceOfTaxonomy( node );
 465         try {
 466             node.getNodeData().getTaxonomy().setTaxonomyCode( tc );
 467         }
 468         catch ( final PhyloXmlDataFormatException e ) {
 469             throw new IllegalArgumentException( "illegal format for taxonomy code: " + tc );
 470         }
 471         return new_value; //TODO //FIXME
 472     }
 473
 474     private static String extractIntermediate( final Map<String, String> intermediate_map, final String name ) {
 475         String new_name = null;
 476         if ( PhylogenyDecorator.VERBOSE ) {
 477             System.out.print( name + " => " );
 478         }
 479         if ( intermediate_map.containsKey( name ) ) {
 480             new_name = intermediate_map.get( name );
 481             if ( ForesterUtil.isEmpty( new_name ) ) {
 482                 throw new IllegalArgumentException( "\"" + name + "\" maps to null or empty string in secondary map" );
 483             }
 484         }
 485         else {
 486             throw new IllegalArgumentException( "\"" + name + "\" not found in name secondary map" );
 487         }
 488         if ( PhylogenyDecorator.VERBOSE ) {
 489             System.out.println( new_name + "  " );
 490         }
 491         return new_name;
 492     }
 493
 494     private static String processNameIntelligently( final String name ) {
 495         final String[] s = name.split( " " );
 496         if ( s.length < 2 ) {
 497             return name;
 498         }
 499         else if ( ( s[ 0 ].indexOf( "_" ) > 0 ) && ( s[ 0 ].indexOf( "|" ) > 0 ) ) {
 500             return s[ 0 ];
 501         }
 502         else if ( ( s[ 1 ].indexOf( "_" ) > 0 ) && ( s[ 1 ].indexOf( "|" ) > 0 ) ) {
 503             return s[ 1 ];
 504         }
 505         else if ( ( s[ 0 ].indexOf( "_" ) > 0 ) && ( s[ 0 ].indexOf( "." ) > 0 ) ) {
 506             return s[ 0 ];
 507         }
 508         else if ( ( s[ 1 ].indexOf( "_" ) > 0 ) && ( s[ 1 ].indexOf( "." ) > 0 ) ) {
 509             return s[ 1 ];
 510         }
 511         else if ( s[ 0 ].indexOf( "_" ) > 0 ) {
 512             return s[ 0 ];
 513         }
 514         else if ( s[ 1 ].indexOf( "_" ) > 0 ) {
 515             return s[ 1 ];
 516         }
 517         else {
 518             return s[ 0 ];
 519         }
 520     }
 521
 522     private static String processSimilarTo( final String name ) {
 523         final int i = name.toLowerCase().indexOf( "similar to" );
 524         String similar_to = "";
 525         if ( i >= 0 ) {
 526             similar_to = " similarity=" + name.substring( i + 10 ).trim();
 527         }
 528         final String pi = processNameIntelligently( name );
 529         return pi + similar_to;
 530     }
 531
 532     private static String sanitize( String s ) {
 533         s = s.replace( ' ', '_' );
 534         s = s.replace( '(', '{' );
 535         s = s.replace( ')', '}' );
 536         s = s.replace( '[', '{' );
 537         s = s.replace( ']', '}' );
 538         s = s.replace( ',', '_' );
 539         return s;
 540     }
 541
 542     public static enum FIELD {
 543         NODE_NAME, SEQUENCE_ANNOTATION_DESC, DOMAIN_STRUCTURE, TAXONOMY_CODE, TAXONOMY_SCIENTIFIC_NAME, SEQUENCE_NAME;
 544     }
 545 }