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.Pattern;
  33
  34 import org.forester.archaeopteryx.AptxUtil;
  35 import org.forester.io.parsers.nhx.NHXFormatException;
  36 import org.forester.io.parsers.phyloxml.PhyloXmlDataFormatException;
  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     final private static Pattern NODENAME_SEQNUMBER_TAXDOMAINNUMBER = Pattern
  67                                                                             .compile( "^([a-fA-Z0-9]{1,5})_([A-Z0-9]{2,4}[A-Z])(\\d{1,4})$" );
  68     public final static boolean  SANITIZE                           = false;
  69     public final static boolean  VERBOSE                            = true;
  70     private static final boolean CUT                                = 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 extract_bracketed_tax_code,
 186                                  final boolean picky,
 187                                  final boolean cut_name_after_space,
 188                                  final boolean process_name_intelligently,
 189                                  final boolean process_similar_to,
 190                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 191                                  final boolean trim_after_tilde ) throws IllegalArgumentException, NHXFormatException,
 192             PhyloXmlDataFormatException {
 193         PhylogenyDecorator.decorate( phylogeny,
 194                                      map,
 195                                      field,
 196                                      extract_bracketed_scientific_name,
 197                                      extract_bracketed_tax_code,
 198                                      picky,
 199                                      null,
 200                                      cut_name_after_space,
 201                                      process_name_intelligently,
 202                                      process_similar_to,
 203                                      numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 204                                      trim_after_tilde );
 205     }
 206
 207     /**
 208      *
 209      *
 210      *
 211      * @param phylogeny
 212      * @param map
 213      *            maps names (in phylogeny) to new values if intermediate_map is
 214      *            null otherwise maps intermediate value to new value
 215      * @param field
 216      * @param picky
 217      * @param intermediate_map
 218      *            maps name (in phylogeny) to a intermediate value
 219      * @throws IllegalArgumentException
 220      * @throws PhyloXmlDataFormatException
 221      */
 222     public static void decorate( final Phylogeny phylogeny,
 223                                  final Map<String, String> map,
 224                                  final FIELD field,
 225                                  final boolean extract_bracketed_scientific_name,
 226                                  final boolean extract_bracketed_tax_code,
 227                                  final boolean picky,
 228                                  final Map<String, String> intermediate_map,
 229                                  final boolean cut_name_after_space,
 230                                  final boolean process_name_intelligently,
 231                                  final boolean process_similar_to,
 232                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 233                                  final boolean trim_after_tilde ) throws IllegalArgumentException,
 234             PhyloXmlDataFormatException {
 235         if ( extract_bracketed_scientific_name && ( field == FIELD.TAXONOMY_SCIENTIFIC_NAME ) ) {
 236             throw new IllegalArgumentException( "attempt to extract bracketed scientific name together with data field pointing to scientific name" );
 237         }
 238         for( final PhylogenyNodeIterator iter = phylogeny.iteratorPostorder(); iter.hasNext(); ) {
 239             final PhylogenyNode node = iter.next();
 240             String name = node.getName();
 241             if ( trim_after_tilde && ( name.indexOf( '~' ) > 0 ) ) {
 242                 name = name.substring( 0, name.indexOf( '~' ) );
 243             }
 244             if ( !ForesterUtil.isEmpty( name ) ) {
 245                 if ( intermediate_map != null ) {
 246                     name = PhylogenyDecorator.extractIntermediate( intermediate_map, name );
 247                 }
 248                 // int space_index = name.indexOf( " " );
 249                 //                if ( CUT && space_index > 0 ) {
 250                 //                    int y = name.lastIndexOf( "|" );
 251                 //                    name = name.substring( y + 1, space_index );
 252                 //                }
 253                 //                String new_value = null;
 254                 //                for( String key : map.keySet() ) {
 255                 //                    if ( key.indexOf( name ) >= 0 ) {
 256                 //                        if ( new_value == null ) {
 257                 //                            new_value = map.get( key );
 258                 //                        }
 259                 //                        else {
 260                 //                            System.out.println( name + " is not unique" );
 261                 //                            System.exit( -1 );
 262                 //                        }
 263                 //                    }
 264                 //                }
 265                 // if ( new_value != null ) {
 266                 if ( map.containsKey( name ) || ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 ) ) {
 267                     String new_value = map.get( name );
 268                     int x = 0;
 269                     while ( ( new_value == null ) && ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 )
 270                             && ( x <= numbers_of_chars_allowed_to_remove_if_not_found_in_map ) ) {
 271                         new_value = map.get( name.substring( 0, name.length() - x ) );
 272                         ++x;
 273                     }
 274                     if ( new_value != null ) {
 275                         new_value = new_value.trim();
 276                         new_value.replaceAll( "/\\s+/", " " );
 277                         if ( extract_bracketed_scientific_name && new_value.endsWith( "]" ) ) {
 278                             new_value = extractBracketedScientificNames( node, new_value );
 279                         }
 280                         else if ( extract_bracketed_tax_code && new_value.endsWith( "]" ) ) {
 281                             new_value = extractBracketedTaxCodes( node, new_value );
 282                         }
 283                         switch ( field ) {
 284                             case SEQUENCE_ANNOTATION_DESC:
 285                                 if ( PhylogenyDecorator.VERBOSE ) {
 286                                     System.out.println( name + ": " + new_value );
 287                                 }
 288                                 if ( !node.getNodeData().isHasSequence() ) {
 289                                     node.getNodeData().setSequence( new Sequence() );
 290                                 }
 291                                 final Annotation annotation = new Annotation( "?" );
 292                                 annotation.setDesc( new_value );
 293                                 node.getNodeData().getSequence().addAnnotation( annotation );
 294                                 break;
 295                             case DOMAIN_STRUCTURE:
 296                                 if ( PhylogenyDecorator.VERBOSE ) {
 297                                     System.out.println( name + ": " + new_value );
 298                                 }
 299                                 if ( !node.getNodeData().isHasSequence() ) {
 300                                     node.getNodeData().setSequence( new Sequence() );
 301                                 }
 302                                 node.getNodeData().getSequence()
 303                                         .setDomainArchitecture( new DomainArchitecture( new_value ) );
 304                                 break;
 305                             case TAXONOMY_CODE:
 306                                 if ( PhylogenyDecorator.VERBOSE ) {
 307                                     System.out.println( name + ": " + new_value );
 308                                 }
 309                                 AptxUtil.ensurePresenceOfTaxonomy( node );
 310                                 node.getNodeData().getTaxonomy().setTaxonomyCode( new_value );
 311                                 break;
 312                             case TAXONOMY_SCIENTIFIC_NAME:
 313                                 if ( PhylogenyDecorator.VERBOSE ) {
 314                                     System.out.println( name + ": " + new_value );
 315                                 }
 316                                 AptxUtil.ensurePresenceOfTaxonomy( node );
 317                                 node.getNodeData().getTaxonomy().setScientificName( new_value );
 318                                 break;
 319                             case SEQUENCE_NAME:
 320                                 if ( PhylogenyDecorator.VERBOSE ) {
 321                                     System.out.println( name + ": " + new_value );
 322                                 }
 323                                 if ( !node.getNodeData().isHasSequence() ) {
 324                                     node.getNodeData().setSequence( new Sequence() );
 325                                 }
 326                                 node.getNodeData().getSequence().setName( new_value );
 327                                 break;
 328                             case NODE_NAME:
 329                                 if ( PhylogenyDecorator.VERBOSE ) {
 330                                     System.out.print( name + " -> " );
 331                                 }
 332                                 if ( cut_name_after_space ) {
 333                                     if ( PhylogenyDecorator.VERBOSE ) {
 334                                         System.out.print( new_value + " -> " );
 335                                     }
 336                                     new_value = PhylogenyDecorator.deleteAtFirstSpace( new_value );
 337                                 }
 338                                 else if ( process_name_intelligently ) {
 339                                     if ( PhylogenyDecorator.VERBOSE ) {
 340                                         System.out.print( new_value + " -> " );
 341                                     }
 342                                     new_value = PhylogenyDecorator.processNameIntelligently( new_value );
 343                                 }
 344                                 else if ( process_similar_to ) {
 345                                     if ( PhylogenyDecorator.VERBOSE ) {
 346                                         System.out.print( new_value + " -> " );
 347                                     }
 348                                     new_value = PhylogenyDecorator.processSimilarTo( new_value );
 349                                 }
 350                                 if ( PhylogenyDecorator.SANITIZE ) {
 351                                     new_value = PhylogenyDecorator.sanitize( new_value );
 352                                 }
 353                                 if ( PhylogenyDecorator.VERBOSE ) {
 354                                     System.out.println( new_value );
 355                                 }
 356                                 node.setName( new_value );
 357                                 break;
 358                             default:
 359                                 throw new RuntimeException( "unknown field \"" + field + "\"" );
 360                         }
 361                     }
 362                 }
 363                 else if ( picky ) {
 364                     throw new IllegalArgumentException( "\"" + name + "\" not found in name map" );
 365                 }
 366             }
 367         }
 368     }
 369
 370     public static void decorate( final Phylogeny[] phylogenies,
 371                                  final Map<String, Map<String, String>> map,
 372                                  final boolean picky,
 373                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map )
 374             throws IllegalArgumentException, NHXFormatException, PhyloXmlDataFormatException {
 375         for( int i = 0; i < phylogenies.length; ++i ) {
 376             PhylogenyDecorator.decorate( phylogenies[ i ],
 377                                          map,
 378                                          picky,
 379                                          numbers_of_chars_allowed_to_remove_if_not_found_in_map );
 380         }
 381     }
 382
 383     public static void decorate( final Phylogeny[] phylogenies,
 384                                  final Map<String, String> map,
 385                                  final FIELD field,
 386                                  final boolean extract_bracketed_scientific_name,
 387                                  final boolean extract_bracketed_tax_code,
 388                                  final boolean picky,
 389                                  final boolean cut_name_after_space,
 390                                  final boolean process_name_intelligently,
 391                                  final boolean process_similar_to,
 392                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 393                                  final boolean trim_after_tilde ) throws IllegalArgumentException, NHXFormatException,
 394             PhyloXmlDataFormatException {
 395         for( int i = 0; i < phylogenies.length; ++i ) {
 396             PhylogenyDecorator.decorate( phylogenies[ i ],
 397                                          map,
 398                                          field,
 399                                          extract_bracketed_scientific_name,
 400                                          extract_bracketed_tax_code,
 401                                          picky,
 402                                          cut_name_after_space,
 403                                          process_name_intelligently,
 404                                          process_similar_to,
 405                                          numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 406                                          trim_after_tilde );
 407         }
 408     }
 409
 410     public static void decorate( final Phylogeny[] phylogenies,
 411                                  final Map<String, String> map,
 412                                  final FIELD field,
 413                                  final boolean extract_bracketed_scientific_name,
 414                                  final boolean extract_bracketed_tax_code,
 415                                  final boolean picky,
 416                                  final Map<String, String> intermediate_map,
 417                                  final boolean cut_name_after_space,
 418                                  final boolean process_name_intelligently,
 419                                  final boolean process_similar_to,
 420                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 421                                  final boolean trim_after_tilde ) throws IllegalArgumentException, NHXFormatException,
 422             PhyloXmlDataFormatException {
 423         for( int i = 0; i < phylogenies.length; ++i ) {
 424             PhylogenyDecorator.decorate( phylogenies[ i ],
 425                                          map,
 426                                          field,
 427                                          extract_bracketed_scientific_name,
 428                                          extract_bracketed_tax_code,
 429                                          picky,
 430                                          intermediate_map,
 431                                          cut_name_after_space,
 432                                          process_name_intelligently,
 433                                          process_similar_to,
 434                                          numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 435                                          trim_after_tilde );
 436         }
 437     }
 438
 439     private static String deleteAtFirstSpace( final String name ) {
 440         final int first_space = name.indexOf( " " );
 441         if ( first_space > 1 ) {
 442             return name.substring( 0, first_space ).trim();
 443         }
 444         return name;
 445     }
 446
 447     private static String extractBracketedScientificNames( final PhylogenyNode node, final String new_value ) {
 448         final int i = new_value.lastIndexOf( "[" );
 449         final String scientific_name = new_value.substring( i + 1, new_value.length() - 1 );
 450         AptxUtil.ensurePresenceOfTaxonomy( node );
 451         node.getNodeData().getTaxonomy().setScientificName( scientific_name );
 452         return new_value.substring( 0, i - 1 ).trim();
 453     }
 454
 455     private static String extractBracketedTaxCodes( final PhylogenyNode node, final String new_value ) {
 456         final int i = new_value.lastIndexOf( "[" );
 457         final String tc = new_value.substring( i + 1, new_value.length() - 1 );
 458         AptxUtil.ensurePresenceOfTaxonomy( node );
 459         try {
 460             node.getNodeData().getTaxonomy().setTaxonomyCode( tc );
 461         }
 462         catch ( final PhyloXmlDataFormatException e ) {
 463             throw new IllegalArgumentException( "illegal format for taxonomy code: " + tc );
 464         }
 465         return new_value.substring( 0, i - 1 ).trim();
 466     }
 467
 468     private static String extractIntermediate( final Map<String, String> intermediate_map, final String name ) {
 469         String new_name = null;
 470         if ( PhylogenyDecorator.VERBOSE ) {
 471             System.out.print( name + " => " );
 472         }
 473         if ( intermediate_map.containsKey( name ) ) {
 474             new_name = intermediate_map.get( name );
 475             if ( ForesterUtil.isEmpty( new_name ) ) {
 476                 throw new IllegalArgumentException( "\"" + name + "\" maps to null or empty string in secondary map" );
 477             }
 478         }
 479         else {
 480             throw new IllegalArgumentException( "\"" + name + "\" not found in name secondary map" );
 481         }
 482         if ( PhylogenyDecorator.VERBOSE ) {
 483             System.out.println( new_name + "  " );
 484         }
 485         return new_name;
 486     }
 487
 488     public static Map<String, Map<String, String>> parseMappingTable( final File mapping_table_file )
 489             throws IOException {
 490         final Map<String, Map<String, String>> map = new HashMap<String, Map<String, String>>();
 491         BasicTable<String> mapping_table = null;
 492         mapping_table = BasicTableParser.parse( mapping_table_file, "\t", false, false );
 493         for( int row = 0; row < mapping_table.getNumberOfRows(); ++row ) {
 494             final Map<String, String> row_map = new HashMap<String, String>();
 495             String name = null;
 496             for( int col = 0; col < mapping_table.getNumberOfColumns(); ++col ) {
 497                 final String table_cell = mapping_table.getValue( col, row );
 498                 if ( col == 0 ) {
 499                     name = table_cell;
 500                 }
 501                 else if ( table_cell != null ) {
 502                     final String key = table_cell.substring( 0, table_cell.indexOf( ':' ) );
 503                     final String val = table_cell.substring( table_cell.indexOf( ':' ) + 1, table_cell.length() );
 504                     row_map.put( key, val );
 505                 }
 506             }
 507             map.put( name, row_map );
 508         }
 509         return map;
 510     }
 511
 512     private static String processNameIntelligently( final String name ) {
 513         final String[] s = name.split( " " );
 514         if ( s.length < 2 ) {
 515             return name;
 516         }
 517         else if ( ( s[ 0 ].indexOf( "_" ) > 0 ) && ( s[ 0 ].indexOf( "|" ) > 0 ) ) {
 518             return s[ 0 ];
 519         }
 520         else if ( ( s[ 1 ].indexOf( "_" ) > 0 ) && ( s[ 1 ].indexOf( "|" ) > 0 ) ) {
 521             return s[ 1 ];
 522         }
 523         else if ( ( s[ 0 ].indexOf( "_" ) > 0 ) && ( s[ 0 ].indexOf( "." ) > 0 ) ) {
 524             return s[ 0 ];
 525         }
 526         else if ( ( s[ 1 ].indexOf( "_" ) > 0 ) && ( s[ 1 ].indexOf( "." ) > 0 ) ) {
 527             return s[ 1 ];
 528         }
 529         else if ( s[ 0 ].indexOf( "_" ) > 0 ) {
 530             return s[ 0 ];
 531         }
 532         else if ( s[ 1 ].indexOf( "_" ) > 0 ) {
 533             return s[ 1 ];
 534         }
 535         else {
 536             return s[ 0 ];
 537         }
 538     }
 539
 540     private static String processSimilarTo( final String name ) {
 541         final int i = name.toLowerCase().indexOf( "similar to" );
 542         String similar_to = "";
 543         if ( i >= 0 ) {
 544             similar_to = " similarity=" + name.substring( i + 10 ).trim();
 545         }
 546         final String pi = processNameIntelligently( name );
 547         return pi + similar_to;
 548     }
 549
 550     private static String sanitize( String s ) {
 551         s = s.replace( ' ', '_' );
 552         s = s.replace( '(', '{' );
 553         s = s.replace( ')', '}' );
 554         s = s.replace( '[', '{' );
 555         s = s.replace( ']', '}' );
 556         s = s.replace( ',', '_' );
 557         return s;
 558     }
 559
 560     public static enum FIELD {
 561         NODE_NAME, SEQUENCE_ANNOTATION_DESC, DOMAIN_STRUCTURE, TAXONOMY_CODE, TAXONOMY_SCIENTIFIC_NAME, SEQUENCE_NAME;
 562     }
 563 }