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.janelia.org";
  49     public static final String TOL_NAME       = "Tree of Life";
  50     public static final String TOL_WEBSERVER  = "http://tolweb.org/onlinecontributors/app?service=external&page=xml/TreeStructureService&node_id="
  51                                                       + PhylogeniesWebserviceClient.QUERY_PLACEHOLDER;
  52     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.";
  53     public static final String TREE_BASE_INST = "treebase";
  54     public static final String TREE_BASE_NAME = "TreeBASE";
  55     public static final String TREE_FAM_INST  = "tree_fam";
  56     public static final String TREE_FAM_NAME  = "TreeFam";
  57
  58     public static List<PhylogeniesWebserviceClient> createDefaultClients() {
  59         final List<PhylogeniesWebserviceClient> clients = new ArrayList<PhylogeniesWebserviceClient>();
  60         clients.add( new BasicPhylogeniesWebserviceClient( TOL_NAME,
  61                                                            "Read Tree from Tree of Life...",
  62                                                            "Use ToL webservice to obtain a evolutionary tree",
  63                                                            "Please enter a Tree of Life node identifier\n(Examples: "
  64                                                                    + "19386 for Cephalopoda, 2461 for Cnidaria, 2466 for Deuterostomia)",
  65                                                            WsPhylogenyFormat.TOL_XML_RESPONSE,
  66                                                            PhylogenyMethods.PhylogenyNodeField.TAXONOMY_SCIENTIFIC_NAME,
  67                                                            WebserviceUtil.TOL_WEBSERVER,
  68                                                            true,
  69                                                            "http://tolweb.org",
  70                                                            null ) );
  71         clients.add( new BasicPhylogeniesWebserviceClient( TREE_BASE_NAME,
  72                                                            "Read Tree(s) from TreeBASE Study...",
  73                                                            "Use TreeBASE to obtain evolutionary tree(s) from a study",
  74                                                            "Please enter a TreeBASE study (\"S\") identifier (without the \"S\")\n(Examples: 15613, 15632, 14525, 14909)",
  75                                                            WsPhylogenyFormat.TREEBASE_STUDY,
  76                                                            null,
  77                                                            "http://purl.org/phylo/treebase/phylows/study/TB2:S"
  78                                                                    + PhylogeniesWebserviceClient.QUERY_PLACEHOLDER
  79                                                                    + "?format=nexus",
  80                                                            true,
  81                                                            "http://www.treebase.org",
  82                                                            TREE_BASE_INST ) );
  83         clients.add( new BasicPhylogeniesWebserviceClient( TREE_BASE_NAME,
  84                                                            "Read Tree from TreeBASE...",
  85                                                            "Use TreeBASE to obtain a evolutionary tree",
  86                                                            "Please enter a TreeBASE tree (\"Tr\") identifier (without the \"Tr\")\n(Examples: 422, 2654, 825, 4931, 2518, 2406, 4934)",
  87                                                            WsPhylogenyFormat.TREEBASE_TREE,
  88                                                            null,
  89                                                            "http://purl.org/phylo/treebase/phylows/tree/TB2:Tr"
  90                                                                    + PhylogeniesWebserviceClient.QUERY_PLACEHOLDER
  91                                                                    + "?format=nexus",
  92                                                            true,
  93                                                            "http://www.treebase.org",
  94                                                            TREE_BASE_INST ) );
  95         clients.add( new BasicPhylogeniesWebserviceClient( PFAM_NAME,
  96                                                            "Read Gene Tree from Pfam...",
  97                                                            "Use  Pfam to obtain gene trees for seed alignments",
  98                                                            "Please enter a Pfam (PF) accession number\n(Examples: 01849 for NAC, 00452 for Bcl-2, 00046 for Homeobox)",
  99                                                            WsPhylogenyFormat.PFAM,
 100                                                            null,
 101                                                            PFAM_SERVER + "/family/PF"
 102                                                                    + PhylogeniesWebserviceClient.QUERY_PLACEHOLDER
 103                                                                    + "/tree/download",
 104                                                            false,
 105                                                            PFAM_SERVER,
 106                                                            PFAM_INST ) );
 107         clients.add( new BasicPhylogeniesWebserviceClient( TREE_FAM_NAME,
 108                                                            "Read Gene Tree from TreeFam...",
 109                                                            "Use TreeFam to obtain a gene tree",
 110                                                            "Please enter a TreeFam (TF) accession number\n(Examples: 101004 for Cyclin D, 315938 for Hox, 105310 for Wnt)",
 111                                                            WsPhylogenyFormat.NHX,
 112                                                            null,
 113                                                            "http://www.treefam.org/family/TF"
 114                                                                    + PhylogeniesWebserviceClient.QUERY_PLACEHOLDER
 115                                                                    + "/tree/newick",
 116                                                            true,
 117                                                            "http://www.treefam.org",
 118                                                            TREE_FAM_INST ) );
 119         return clients;
 120     }
 121
 122     public static void processInstructions( final PhylogeniesWebserviceClient client, final Phylogeny phylogeny )
 123             throws PhyloXmlDataFormatException {
 124         if ( client.getProcessingInstructions().equals( WebserviceUtil.TREE_FAM_INST ) ) {
 125             WebserviceUtil.processTreeFamTrees( phylogeny );
 126         }
 127         else if ( client.getProcessingInstructions().equals( WebserviceUtil.PFAM_INST ) ) {
 128             WebserviceUtil.extractSpTremblAccFromNodeName( phylogeny, "sptrembl" );
 129             PhylogenyMethods.transferInternalNodeNamesToConfidence( phylogeny, "bootstrap" );
 130         }
 131         else if ( client.getProcessingInstructions().equals( WebserviceUtil.TREE_BASE_INST ) ) {
 132             if ( PhylogenyMethods.isInternalNamesLookLikeConfidences( phylogeny ) ) {
 133                 PhylogenyMethods.transferInternalNodeNamesToConfidence( phylogeny, "" );
 134             }
 135             WebserviceUtil.processTreeBaseTrees( phylogeny );
 136         }
 137     }
 138
 139     static void extractSpTremblAccFromNodeName( final Phylogeny phy, final String source ) {
 140         final PreorderTreeIterator it = new PreorderTreeIterator( phy );
 141         while ( it.hasNext() ) {
 142             final PhylogenyNode n = it.next();
 143             if ( !ForesterUtil.isEmpty( n.getName() ) ) {
 144                 final String name = n.getName();
 145                 final int i = name.lastIndexOf( "/" );
 146                 if ( i > 0 ) {
 147                     final String acc_str = name.substring( 0, i );
 148                     if ( !ForesterUtil.isEmpty( acc_str ) ) {
 149                         final Sequence seq = new Sequence();
 150                         final Accession acc = new Accession( acc_str, source );
 151                         seq.setAccession( acc );
 152                         n.getNodeData().setSequence( seq );
 153                     }
 154                 }
 155             }
 156         }
 157     }
 158
 159     static void processTreeBaseTrees( final Phylogeny phy ) {
 160         phy.setDescription( TREE_BASE_DESC );
 161         final PhylogenyNodeIterator it = phy.iteratorExternalForward();
 162         while ( it.hasNext() ) {
 163             final PhylogenyNode n = it.next();
 164             if ( !ForesterUtil.isEmpty( n.getName() ) ) {
 165                 final Accession acc = SequenceAccessionTools.parseAccessorFromString( n.getName() );
 166                 if ( acc != null ) {
 167                     if ( !n.getNodeData().isHasSequence() ) {
 168                         n.getNodeData().addSequence( new Sequence() );
 169                     }
 170                     final Sequence s = n.getNodeData().getSequence();
 171                     if ( s.getAccession() == null ) {
 172                         s.setAccession( acc );
 173                     }
 174                 }
 175             }
 176         }
 177     }
 178
 179     static void processTreeFamTrees( final Phylogeny phy ) {
 180         final PhylogenyNodeIterator it = phy.iteratorPostorder();
 181         while ( it.hasNext() ) {
 182             final PhylogenyNode n = it.next();
 183             if ( n.isExternal() ) {
 184                 n.getNodeData().setEvent( null );
 185                 if ( !ForesterUtil.isEmpty( n.getName() ) ) {
 186                     final Accession acc = SequenceAccessionTools.parseAccessorFromString( n.getName() );
 187                     if ( acc != null ) {
 188                         if ( !n.getNodeData().isHasSequence() ) {
 189                             n.getNodeData().addSequence( new Sequence() );
 190                         }
 191                         final Sequence s = n.getNodeData().getSequence();
 192                         if ( s.getAccession() == null ) {
 193                             s.setAccession( acc );
 194                         }
 195                     }
 196                 }
 197             }
 198             else {
 199                 if ( ( n.getBranchData() != null ) && n.getBranchData().isHasConfidences()
 200                         && ( n.getBranchData().getConfidence( 0 ) != null ) ) {
 201                     n.getBranchData().getConfidence( 0 ).setType( "bootstrap" );
 202                 }
 203                 if ( !ForesterUtil.isEmpty( n.getName() ) ) {
 204                     if ( !n.getNodeData().isHasTaxonomy() ) {
 205                         n.getNodeData().addTaxonomy( new Taxonomy() );
 206                     }
 207                     final Taxonomy t = n.getNodeData().getTaxonomy();
 208                     if ( ForesterUtil.isEmpty( t.getScientificName() ) ) {
 209                         t.setScientificName( n.getName() );
 210                         n.setName( "" );
 211                     }
 212                 }
 213             }
 214             if ( n.getNodeData().isHasTaxonomy() && ( n.getNodeData().getTaxonomy().getIdentifier() != null ) ) {
 215                 n.getNodeData()
 216                         .getTaxonomy()
 217                         .setIdentifier( new Identifier( n.getNodeData().getTaxonomy().getIdentifier().getValue(),
 218                                                         "ncbi" ) );
 219             }
 220         }
 221     }
 222 }