forester/java/src/org/forester/archaeopteryx/webservices/WebserviceUtil.java

   1 // $Id:
   2 // forester -- software libraries and applications
   3 // for evolutionary biology research and applications.
   4 //
   5 // Copyright (C) 2008-2010 Christian M. Zmasek
   6 // Copyright (C) 2008-2010 Burnham Institute for Medical Research
   7 //
   8 // This library is free software; you can redistribute it and/or
   9 // modify it under the terms of the GNU Lesser General Public
  10 // License as published by the Free Software Foundation; either
  11 // version 2.1 of the License, or (at your option) any later version.
  12 //
  13 // This library is distributed in the hope that it will be useful,
  14 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  15 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  16 // Lesser General Public License for more details.
  17 //
  18 // You should have received a copy of the GNU Lesser General Public
  19 // License along with this library; if not, write to the Free Software
  20 // Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
  21 //
  22 // Contact: phylosoft @ gmail . com
  23 // WWW: https://sites.google.com/site/cmzmasek/home/software/forester
  24
  25 package org.forester.archaeopteryx.webservices;
  26
  27 import java.util.ArrayList;
  28 import java.util.List;
  29
  30 import org.forester.archaeopteryx.webservices.WebservicesManager.WsPhylogenyFormat;
  31 import org.forester.io.parsers.phyloxml.PhyloXmlDataFormatException;
  32 import org.forester.phylogeny.Phylogeny;
  33 import org.forester.phylogeny.PhylogenyMethods;
  34 import org.forester.phylogeny.PhylogenyNode;
  35 import org.forester.phylogeny.data.Accession;
  36 import org.forester.phylogeny.data.Identifier;
  37 import org.forester.phylogeny.data.Sequence;
  38 import org.forester.phylogeny.data.Taxonomy;
  39 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
  40 import org.forester.phylogeny.iterators.PreorderTreeIterator;
  41 import org.forester.util.ForesterUtil;
  42 import org.forester.util.SequenceAccessionTools;
  43
  44 public final class WebserviceUtil {
  45
  46     public static final String PFAM_INST                       = "pfam";
  47     public static final String PFAM_NAME                       = "Pfam";
  48     public static final String PFAM_SERVER                     = "http://pfam.xfam.org";
  49     public static final String TOL_NAME                        = "Tree of Life (ToL)";
  50     public static final String TOL_URL_BASE                    = "http://tolweb.org/onlinecontributors/app?service=external&page=xml/TreeStructureService&node_id=";
  51     public static final String TOL_WEBSERVER                   = TOL_URL_BASE
  52             + PhylogeniesWebserviceClient.QUERY_PLACEHOLDER;
  53     public static final String TREE_BASE_DESC                  = "This data set was downloaded from TreeBASE, a relational database of phylogenetic knowledge. TreeBASE has been supported by the NSF, Harvard University, Yale University, SDSC and UC Davis. Please do not remove this acknowledgment.";
  54     public static final String TREE_BASE_INST                  = "treebase";
  55     public static final String TREE_BASE_NAME                  = "TreeBASE";
  56     public static final String TREE_FAM_INST                   = "tree_fam";
  57     public static final String TREE_FAM_NAME                   = "TreeFam";
  58     public static final String TREE_FAM_URL_BASE               = "http://www.treefam.org/family/TF";
  59     public static final String TREEBASE_PHYLOWS_STUDY_URL_BASE = "http://purl.org/phylo/treebase/phylows/study/TB2:S";
  60   //  public static final String TREEBASE_PHYLOWS_TREE_URL_BASE  = "http://purl.org/phylo/treebase/phylows/tree/TB2:Tr";
  61     public static final String TREEBASE_PHYLOWS_TREE_URL_BASE  = "https://treebase.org/treebase-web/phylows/tree/TB2:Tr";
  62
  63
  64
  65     public static List<PhylogeniesWebserviceClient> createDefaultClients() {
  66         final List<PhylogeniesWebserviceClient> clients = new ArrayList<PhylogeniesWebserviceClient>();
  67         clients.add( new BasicPhylogeniesWebserviceClient( TREE_BASE_NAME,
  68                                                            "Read Tree(s) from TreeBASE Study...",
  69                                                            "Use TreeBASE to obtain evolutionary tree(s) from a study",
  70                                                            "Please enter a TreeBASE study (\"S\") identifier (without the \"S\")\n(Examples: 14909, 14525, 15613, 15632)",
  71                                                            WsPhylogenyFormat.TREEBASE_STUDY,
  72                                                            null,
  73                                                            TREEBASE_PHYLOWS_STUDY_URL_BASE
  74                                                            + PhylogeniesWebserviceClient.QUERY_PLACEHOLDER
  75                                                            + "?format=nexus",
  76                                                            true,
  77                                                            "http://www.treebase.org",
  78                                                            TREE_BASE_INST ) );
  79         clients.add( new BasicPhylogeniesWebserviceClient( TREE_BASE_NAME,
  80                                                            "Read Tree from TreeBASE...",
  81                                                            "Use TreeBASE to obtain a evolutionary tree",
  82                                                            "Please enter a TreeBASE tree (\"Tr\") identifier (without the \"Tr\")\n(Examples: 2406, 422, 2654, 825, 4931, 2518, 4934)",
  83                                                            WsPhylogenyFormat.TREEBASE_TREE,
  84                                                            null,
  85                                                            TREEBASE_PHYLOWS_TREE_URL_BASE
  86                                                            + PhylogeniesWebserviceClient.QUERY_PLACEHOLDER
  87                                                            + "?format=nexus",
  88                                                            true,
  89                                                            "http://www.treebase.org",
  90                                                            TREE_BASE_INST ) );
  91         clients.add( new BasicPhylogeniesWebserviceClient( PFAM_NAME,
  92                                                            "Read Domain Tree from Pfam...",
  93                                                            "Use  Pfam to obtain gene trees for seed alignments",
  94                                                            "Please enter a Pfam (PF) accession number\n(Examples: 01849 for NAC, 00452 for Bcl-2, 00046 for Homeobox)",
  95                                                            WsPhylogenyFormat.PFAM,
  96                                                            null,
  97                                                            PFAM_SERVER + "/family/PF"
  98                                                                    + PhylogeniesWebserviceClient.QUERY_PLACEHOLDER
  99                                                                    + "/tree/download",
 100                                                                    false,
 101                                                                    PFAM_SERVER,
 102                                                                    PFAM_INST ) );
 103         clients.add( new BasicPhylogeniesWebserviceClient( TREE_FAM_NAME,
 104                                                            "Read Gene Tree from TreeFam...",
 105                                                            "Use TreeFam to obtain a gene tree",
 106                                                            "Please enter a TreeFam (TF) accession number\n(Examples: 101004 for Cyclin D, 315938 for Hox, 105310 for Wnt)",
 107                                                            WsPhylogenyFormat.NHX,
 108                                                            null,
 109                                                            TREE_FAM_URL_BASE
 110                                                            + PhylogeniesWebserviceClient.QUERY_PLACEHOLDER
 111                                                            + "/tree/newick",
 112                                                            true,
 113                                                            "http://www.treefam.org",
 114                                                            TREE_FAM_INST ) );
 115         clients.add( new BasicPhylogeniesWebserviceClient( TOL_NAME,
 116                                                            "Read Tree from Tree of Life (ToL)...",
 117                                                            "Use ToL webservice to obtain a evolutionary tree",
 118                                                            "Please enter a Tree of Life node identifier\n(Examples: "
 119                                                                    + "14923 for ray-finned fishes, 19386 for Cephalopoda, 2461 for Cnidaria)",
 120                                                                    WsPhylogenyFormat.TOL_XML_RESPONSE,
 121                                                                    PhylogenyMethods.PhylogenyNodeField.TAXONOMY_SCIENTIFIC_NAME,
 122                                                                    WebserviceUtil.TOL_WEBSERVER,
 123                                                                    true,
 124                                                                    "http://tolweb.org",
 125                                                                    null ) );
 126         return clients;
 127     }
 128
 129     public static void processInstructions( final PhylogeniesWebserviceClient client, final Phylogeny phylogeny )
 130             throws PhyloXmlDataFormatException {
 131         if ( client.getProcessingInstructions().equals( WebserviceUtil.TREE_FAM_INST ) ) {
 132             WebserviceUtil.processTreeFamTrees( phylogeny );
 133         }
 134         else if ( client.getProcessingInstructions().equals( WebserviceUtil.PFAM_INST ) ) {
 135             WebserviceUtil.extractSpTremblAccFromNodeName( phylogeny, "sptrembl" );
 136             PhylogenyMethods.transferInternalNodeNamesToConfidence( phylogeny, "bootstrap" );
 137         }
 138         else if ( client.getProcessingInstructions().equals( WebserviceUtil.TREE_BASE_INST ) ) {
 139             if ( PhylogenyMethods.isInternalNamesLookLikeConfidences( phylogeny ) ) {
 140                 PhylogenyMethods.transferInternalNodeNamesToConfidence( phylogeny, "" );
 141             }
 142             WebserviceUtil.processTreeBaseTrees( phylogeny );
 143         }
 144     }
 145
 146     static void extractSpTremblAccFromNodeName( final Phylogeny phy, final String source ) {
 147         final PreorderTreeIterator it = new PreorderTreeIterator( phy );
 148         while ( it.hasNext() ) {
 149             final PhylogenyNode n = it.next();
 150             if ( !ForesterUtil.isEmpty( n.getName() ) ) {
 151                 final String name = n.getName();
 152                 final int i = name.lastIndexOf( "/" );
 153                 if ( i > 0 ) {
 154                     final String acc_str = name.substring( 0, i );
 155                     if ( !ForesterUtil.isEmpty( acc_str ) ) {
 156                         final Sequence seq = new Sequence();
 157                         final Accession acc = new Accession( acc_str, source );
 158                         seq.setAccession( acc );
 159                         n.getNodeData().setSequence( seq );
 160                     }
 161                 }
 162             }
 163         }
 164     }
 165
 166     static void processTreeBaseTrees( final Phylogeny phy ) {
 167         phy.setDescription( TREE_BASE_DESC );
 168         final PhylogenyNodeIterator it = phy.iteratorExternalForward();
 169         while ( it.hasNext() ) {
 170             final PhylogenyNode n = it.next();
 171             if ( !ForesterUtil.isEmpty( n.getName() ) ) {
 172                 final Accession acc = SequenceAccessionTools.parseAccessorFromString( n.getName() );
 173                 if ( acc != null ) {
 174                     if ( !n.getNodeData().isHasSequence() ) {
 175                         n.getNodeData().addSequence( new Sequence() );
 176                     }
 177                     final Sequence s = n.getNodeData().getSequence();
 178                     if ( s.getAccession() == null ) {
 179                         s.setAccession( acc );
 180                     }
 181                 }
 182             }
 183         }
 184     }
 185
 186     static void processTreeFamTrees( final Phylogeny phy ) {
 187         final PhylogenyNodeIterator it = phy.iteratorPostorder();
 188         while ( it.hasNext() ) {
 189             final PhylogenyNode n = it.next();
 190             if ( n.isExternal() ) {
 191                 n.getNodeData().setEvent( null );
 192                 if ( !ForesterUtil.isEmpty( n.getName() ) ) {
 193                     final Accession acc = SequenceAccessionTools.parseAccessorFromString( n.getName() );
 194                     if ( acc != null ) {
 195                         if ( !n.getNodeData().isHasSequence() ) {
 196                             n.getNodeData().addSequence( new Sequence() );
 197                         }
 198                         final Sequence s = n.getNodeData().getSequence();
 199                         if ( s.getAccession() == null ) {
 200                             s.setAccession( acc );
 201                         }
 202                     }
 203                 }
 204             }
 205             else {
 206                 if ( ( n.getBranchData() != null ) && n.getBranchData().isHasConfidences()
 207                         && ( n.getBranchData().getConfidence( 0 ) != null ) ) {
 208                     n.getBranchData().getConfidence( 0 ).setType( "bootstrap" );
 209                 }
 210                 if ( !ForesterUtil.isEmpty( n.getName() ) ) {
 211                     if ( !n.getNodeData().isHasTaxonomy() ) {
 212                         n.getNodeData().addTaxonomy( new Taxonomy() );
 213                     }
 214                     final Taxonomy t = n.getNodeData().getTaxonomy();
 215                     if ( ForesterUtil.isEmpty( t.getScientificName() ) ) {
 216                         t.setScientificName( n.getName() );
 217                         n.setName( "" );
 218                     }
 219                 }
 220             }
 221             if ( n.getNodeData().isHasTaxonomy() && ( n.getNodeData().getTaxonomy().getIdentifier() != null ) ) {
 222                 n.getNodeData()
 223                 .getTaxonomy()
 224                 .setIdentifier( new Identifier( n.getNodeData().getTaxonomy().getIdentifier().getValue(),
 225                                                 "ncbi" ) );
 226             }
 227         }
 228     }
 229 }