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 = "https://treebase.org/treebase-web/phylows/study/TB2:S";
  60     public static final String TREEBASE_PHYLOWS_TREE_URL_BASE  = "https://treebase.org/treebase-web/phylows/tree/TB2:Tr";
  61
  62
  63
  64     public static List<PhylogeniesWebserviceClient> createDefaultClients() {
  65         final List<PhylogeniesWebserviceClient> clients = new ArrayList<PhylogeniesWebserviceClient>();
  66         clients.add( new BasicPhylogeniesWebserviceClient( TREE_BASE_NAME,
  67                                                            "Read Tree(s) from TreeBASE Study...",
  68                                                            "Use TreeBASE to obtain evolutionary tree(s) from a study",
  69                                                            "Please enter a TreeBASE study (\"S\") identifier (without the \"S\")\n(Examples: 13246, 11662, 562, 16424, 17878, 10190, 14909, 14525, 15613, 15632)",
  70                                                            WsPhylogenyFormat.TREEBASE_STUDY,
  71                                                            null,
  72                                                            TREEBASE_PHYLOWS_STUDY_URL_BASE
  73                                                            + PhylogeniesWebserviceClient.QUERY_PLACEHOLDER
  74                                                            + "?format=nexus",
  75                                                            true,
  76                                                            "https://treebase.org",
  77                                                            TREE_BASE_INST ) );
  78         clients.add( new BasicPhylogeniesWebserviceClient( TREE_BASE_NAME,
  79                                                            "Read Tree from TreeBASE...",
  80                                                            "Use TreeBASE to obtain a evolutionary tree",
  81                                                            "Please enter a TreeBASE tree (\"Tr\") identifier (without the \"Tr\")\n(Examples: 2406, 422, 2654, 825, 4931, 2518, 4934)",
  82                                                            WsPhylogenyFormat.TREEBASE_TREE,
  83                                                            null,
  84                                                            TREEBASE_PHYLOWS_TREE_URL_BASE
  85                                                            + PhylogeniesWebserviceClient.QUERY_PLACEHOLDER
  86                                                            + "?format=nexus",
  87                                                            true,
  88                                                            "https://treebase.org",
  89                                                            TREE_BASE_INST ) );
  90         clients.add( new BasicPhylogeniesWebserviceClient( PFAM_NAME,
  91                                                            "Read Domain Tree from Pfam...",
  92                                                            "Use  Pfam to obtain gene trees for seed alignments",
  93                                                            "Please enter a Pfam (PF) accession number\n(Examples: 01849 for NAC, 00452 for Bcl-2, 00046 for Homeobox)",
  94                                                            WsPhylogenyFormat.PFAM,
  95                                                            null,
  96                                                            PFAM_SERVER + "/family/PF"
  97                                                                    + PhylogeniesWebserviceClient.QUERY_PLACEHOLDER
  98                                                                    + "/tree/download",
  99                                                                    false,
 100                                                                    PFAM_SERVER,
 101                                                                    PFAM_INST ) );
 102         clients.add( new BasicPhylogeniesWebserviceClient( TREE_FAM_NAME,
 103                                                            "Read Gene Tree from TreeFam...",
 104                                                            "Use TreeFam to obtain a gene tree",
 105                                                            "Please enter a TreeFam (TF) accession number\n(Examples: 101004 for Cyclin D, 315938 for Hox, 105310 for Wnt)",
 106                                                            WsPhylogenyFormat.NHX,
 107                                                            null,
 108                                                            TREE_FAM_URL_BASE
 109                                                            + PhylogeniesWebserviceClient.QUERY_PLACEHOLDER
 110                                                            + "/tree/newick",
 111                                                            true,
 112                                                            "http://www.treefam.org",
 113                                                            TREE_FAM_INST ) );
 114         clients.add( new BasicPhylogeniesWebserviceClient( TOL_NAME,
 115                                                            "Read Tree from Tree of Life (ToL)...",
 116                                                            "Use ToL webservice to obtain a evolutionary tree",
 117                                                            "Please enter a Tree of Life node identifier\n(Examples: "
 118                                                                    + "14923 for ray-finned fishes, 19386 for Cephalopoda, 2461 for Cnidaria)",
 119                                                                    WsPhylogenyFormat.TOL_XML_RESPONSE,
 120                                                                    PhylogenyMethods.PhylogenyNodeField.TAXONOMY_SCIENTIFIC_NAME,
 121                                                                    WebserviceUtil.TOL_WEBSERVER,
 122                                                                    true,
 123                                                                    "http://tolweb.org",
 124                                                                    null ) );
 125         return clients;
 126     }
 127
 128     public static void processInstructions( final PhylogeniesWebserviceClient client, final Phylogeny phylogeny )
 129             throws PhyloXmlDataFormatException {
 130         if ( client.getProcessingInstructions().equals( WebserviceUtil.TREE_FAM_INST ) ) {
 131             WebserviceUtil.processTreeFamTrees( phylogeny );
 132         }
 133         else if ( client.getProcessingInstructions().equals( WebserviceUtil.PFAM_INST ) ) {
 134             WebserviceUtil.extractSpTremblAccFromNodeName( phylogeny, "sptrembl" );
 135             PhylogenyMethods.transferInternalNodeNamesToConfidence( phylogeny, "bootstrap" );
 136         }
 137         else if ( client.getProcessingInstructions().equals( WebserviceUtil.TREE_BASE_INST ) ) {
 138             if ( PhylogenyMethods.isInternalNamesLookLikeConfidences( phylogeny ) ) {
 139                 PhylogenyMethods.transferInternalNodeNamesToConfidence( phylogeny, "" );
 140             }
 141             WebserviceUtil.processTreeBaseTrees( phylogeny );
 142         }
 143     }
 144
 145     static void extractSpTremblAccFromNodeName( final Phylogeny phy, final String source ) {
 146         final PreorderTreeIterator it = new PreorderTreeIterator( phy );
 147         while ( it.hasNext() ) {
 148             final PhylogenyNode n = it.next();
 149             if ( !ForesterUtil.isEmpty( n.getName() ) ) {
 150                 final String name = n.getName();
 151                 final int i = name.lastIndexOf( "/" );
 152                 if ( i > 0 ) {
 153                     final String acc_str = name.substring( 0, i );
 154                     if ( !ForesterUtil.isEmpty( acc_str ) ) {
 155                         final Sequence seq = new Sequence();
 156                         final Accession acc = new Accession( acc_str, source );
 157                         seq.setAccession( acc );
 158                         n.getNodeData().setSequence( seq );
 159                     }
 160                 }
 161             }
 162         }
 163     }
 164
 165     static void processTreeBaseTrees( final Phylogeny phy ) {
 166         phy.setDescription( TREE_BASE_DESC );
 167         final PhylogenyNodeIterator it = phy.iteratorExternalForward();
 168         while ( it.hasNext() ) {
 169             final PhylogenyNode n = it.next();
 170             if ( !ForesterUtil.isEmpty( n.getName() ) ) {
 171                 final Accession acc = SequenceAccessionTools.parseAccessorFromString( n.getName() );
 172                 if ( acc != null ) {
 173                     if ( !n.getNodeData().isHasSequence() ) {
 174                         n.getNodeData().addSequence( new Sequence() );
 175                     }
 176                     final Sequence s = n.getNodeData().getSequence();
 177                     if ( s.getAccession() == null ) {
 178                         s.setAccession( acc );
 179                     }
 180                 }
 181             }
 182         }
 183     }
 184
 185     static void processTreeFamTrees( final Phylogeny phy ) {
 186         final PhylogenyNodeIterator it = phy.iteratorPostorder();
 187         while ( it.hasNext() ) {
 188             final PhylogenyNode n = it.next();
 189             if ( n.isExternal() ) {
 190                 n.getNodeData().setEvent( null );
 191                 if ( !ForesterUtil.isEmpty( n.getName() ) ) {
 192                     final Accession acc = SequenceAccessionTools.parseAccessorFromString( n.getName() );
 193                     if ( acc != null ) {
 194                         if ( !n.getNodeData().isHasSequence() ) {
 195                             n.getNodeData().addSequence( new Sequence() );
 196                         }
 197                         final Sequence s = n.getNodeData().getSequence();
 198                         if ( s.getAccession() == null ) {
 199                             s.setAccession( acc );
 200                         }
 201                     }
 202                 }
 203             }
 204             else {
 205                 if ( ( n.getBranchData() != null ) && n.getBranchData().isHasConfidences()
 206                         && ( n.getBranchData().getConfidence( 0 ) != null ) ) {
 207                     n.getBranchData().getConfidence( 0 ).setType( "bootstrap" );
 208                 }
 209                 if ( !ForesterUtil.isEmpty( n.getName() ) ) {
 210                     if ( !n.getNodeData().isHasTaxonomy() ) {
 211                         n.getNodeData().addTaxonomy( new Taxonomy() );
 212                     }
 213                     final Taxonomy t = n.getNodeData().getTaxonomy();
 214                     if ( ForesterUtil.isEmpty( t.getScientificName() ) ) {
 215                         t.setScientificName( n.getName() );
 216                         n.setName( "" );
 217                     }
 218                 }
 219             }
 220             if ( n.getNodeData().isHasTaxonomy() && ( n.getNodeData().getTaxonomy().getIdentifier() != null ) ) {
 221                 n.getNodeData()
 222                 .getTaxonomy()
 223                 .setIdentifier( new Identifier( n.getNodeData().getTaxonomy().getIdentifier().getValue(),
 224                                                 "ncbi" ) );
 225             }
 226         }
 227     }
 228 }