forester/java/src/org/forester/ws/uniprot/UniProtTaxonomy.java

   1 // $Id:
   2 // forester -- software libraries and applications
   3 // for genomics and evolutionary biology research.
   4 //
   5 // Copyright (C) 2010 Christian M Zmasek
   6 // Copyright (C) 2010 Sanford-Burnham Medical Research Institute
   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.ws.uniprot;
  27
  28 import java.util.ArrayList;
  29 import java.util.List;
  30
  31 import org.forester.util.ForesterUtil;
  32
  33 public final class UniProtTaxonomy {
  34
  35     private static final String         ARCHAEA                  = "Archaea";
  36     private static final String         BACTERIA                 = "Bacteria";
  37     private static final String         EUKARYOTA                = "Eukaryota";
  38     private final List<String>          _lineage;
  39     private final String                _code;
  40     private final String                _scientific_name;
  41     private final String                _common_name;
  42     private final String                _synonym;
  43     private final String                _rank;
  44     private final String                _id;
  45     public final static String          CELLULAR_ORGANISMS       = "cellular organisms";
  46     public final static String          VIRUSES                  = "Viruses";
  47     public final static UniProtTaxonomy DROSOPHILA_GENUS         = new UniProtTaxonomy( new String[] {
  48             CELLULAR_ORGANISMS, EUKARYOTA, "Metazoa", "Arthropoda", "Hexapoda", "Insecta", "Pterygota", "Neoptera",
  49             "Endopterygota", "Diptera", "Brachycera", "Muscomorpha", "Ephydroidea", "Drosophilidae", "Drosophila" },
  50                                                                                         "",
  51                                                                                         "fruit flies",
  52                                                                                         "Drosophila",
  53                                                                                         "",
  54                                                                                         "genus",
  55                                                                                         "7215" );
  56     public final static UniProtTaxonomy XENOPUS_GENUS            = new UniProtTaxonomy( new String[] {
  57             CELLULAR_ORGANISMS, EUKARYOTA, "Metazoa", "Chordata", "Craniata", "Vertebrata", "Euteleostomi", "Amphibia",
  58             "Batrachia", "Anura", "Mesobatrachia", "Pipoidea", "Pipidae", "Xenopodinae", "Xenopus" },
  59                                                                                         "",
  60                                                                                         "",
  61                                                                                         "Xenopus",
  62                                                                                         "",
  63                                                                                         "genus",
  64                                                                                         "8353" );
  65     public final static UniProtTaxonomy CAPITELLA_TELATA_SPECIES = new UniProtTaxonomy( new String[] {
  66             CELLULAR_ORGANISMS, EUKARYOTA, "Metazoa", "Annelida", "Polychaeta", "Scolecida", "Capitellida",
  67             "Capitellidae", "Capitella", "Capitella teleta"                            },
  68                                                                                         "",
  69                                                                                         "",
  70                                                                                         "Capitella teleta",
  71                                                                                         "Capitella sp. I",
  72                                                                                         "species",
  73                                                                                         "283909" );
  74
  75     // public final static UniProtTaxonomy NUCLEARIIDAE_AND_FONTICULA = new UniProtTaxonomy( new String[] {
  76     //         CELLULAR_ORGANISMS, EUKARYOTA, "Nucleariidae and Fonticula group" }, "", "", "", "", "", "1001604" );
  77     public UniProtTaxonomy( final String line ) {
  78         final String[] items = line.split( "\t" );
  79         if ( items.length < 5 ) {
  80             throw new IllegalArgumentException( "cannot parse uniprot taxonomy from: " + line );
  81         }
  82         _id = items[ 0 ].trim();
  83         _code = items[ 1 ].trim();
  84         _scientific_name = items[ 2 ].trim();
  85         _common_name = items[ 3 ].trim();
  86         _synonym = items[ 4 ].trim();
  87         if ( items.length > 6 ) {
  88             _rank = items[ 7 ].trim();
  89         }
  90         else {
  91             _rank = "";
  92         }
  93         String[] lin = null;
  94         if ( items.length > 8 ) {
  95             lin = items[ 8 ].split( "; " );
  96         }
  97         _lineage = new ArrayList<String>();
  98         if ( ( lin != null ) && ( lin.length > 0 ) ) {
  99             final List<String> temp = new ArrayList<String>();
 100             for( final String t : lin ) {
 101                 if ( !ForesterUtil.isEmpty( t ) ) {
 102                     temp.add( t.trim() );
 103                 }
 104             }
 105             for( int i = 0; i < temp.size(); ++i ) {
 106                 if ( ( i == 0 )
 107                         && ( temp.get( i ).equalsIgnoreCase( EUKARYOTA ) || temp.get( i ).equalsIgnoreCase( BACTERIA ) || temp
 108                                 .get( i ).equalsIgnoreCase( ARCHAEA ) ) ) {
 109                     _lineage.add( CELLULAR_ORGANISMS );
 110                 }
 111                 _lineage.add( temp.get( i ) );
 112             }
 113         }
 114         if ( _lineage.isEmpty()
 115                 && ( _scientific_name.equalsIgnoreCase( EUKARYOTA ) || _scientific_name.equalsIgnoreCase( BACTERIA ) || _scientific_name
 116                         .equalsIgnoreCase( ARCHAEA ) ) ) {
 117             _lineage.add( CELLULAR_ORGANISMS );
 118         }
 119         _lineage.add( _scientific_name );
 120         if ( _lineage.isEmpty() ) {
 121             throw new IllegalArgumentException( "lineage in a UniProt Taxonomy can not be empty\n: " + line );
 122         }
 123     }
 124
 125     public UniProtTaxonomy( final List<String> lineage,
 126                             final String code,
 127                             final String common_name,
 128                             final String scientific_name,
 129                             final String synonym,
 130                             final String rank,
 131                             final String id ) {
 132         _lineage = lineage;
 133         _code = code;
 134         _scientific_name = scientific_name;
 135         _common_name = common_name;
 136         _synonym = synonym;
 137         _rank = rank;
 138         _id = id;
 139         if ( ( _lineage != null ) && !_lineage.get( _lineage.size() - 1 ).equalsIgnoreCase( _scientific_name ) ) {
 140             _lineage.add( _scientific_name );
 141         }
 142     }
 143
 144     public UniProtTaxonomy( final String[] lineage,
 145                             final String code,
 146                             final String common_name,
 147                             final String scientific_name,
 148                             final String synonym,
 149                             final String rank,
 150                             final String id ) {
 151         _lineage = new ArrayList<String>();
 152         if ( lineage != null ) {
 153             for( final String l : lineage ) {
 154                 _lineage.add( l );
 155             }
 156         }
 157         _code = code;
 158         _scientific_name = scientific_name;
 159         _common_name = common_name;
 160         _synonym = synonym;
 161         _rank = rank;
 162         _id = id;
 163         if ( ( _lineage != null ) && !_lineage.get( _lineage.size() - 1 ).equalsIgnoreCase( _scientific_name ) ) {
 164             _lineage.add( _scientific_name );
 165         }
 166     }
 167
 168     /**
 169      * Creates deep copy for all fields, except lineage.
 170      *
 171      * @return
 172      */
 173     public UniProtTaxonomy copy() {
 174         return new UniProtTaxonomy( getLineage(),
 175                                     getCode() != null ? new String( getCode() ) : null,
 176                                     getCommonName() != null ? new String( getCommonName() ) : null,
 177                                     getScientificName() != null ? new String( getScientificName() ) : null,
 178                                     getSynonym() != null ? new String( getSynonym() ) : null,
 179                                     getRank() != null ? new String( getRank() ) : null,
 180                                     getId() != null ? new String( getId() ) : null );
 181     }
 182
 183     public String getCode() {
 184         return _code;
 185     }
 186
 187     public String getCommonName() {
 188         return _common_name;
 189     }
 190
 191     public String getId() {
 192         return _id;
 193     }
 194
 195     public List<String> getLineage() {
 196         return _lineage;
 197     }
 198
 199     public String getRank() {
 200         return _rank;
 201     }
 202
 203     public String getScientificName() {
 204         return _scientific_name;
 205     }
 206
 207     public String getSynonym() {
 208         return _synonym;
 209     }
 210 }