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: www.phylosoft.org/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.PhyloXmlUtil;
  32 import org.forester.phylogeny.Phylogeny;
  33 import org.forester.phylogeny.PhylogenyNode;
  34 import org.forester.phylogeny.data.Accession;
  35 import org.forester.phylogeny.data.Identifier;
  36 import org.forester.phylogeny.data.Sequence;
  37 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
  38 import org.forester.phylogeny.iterators.PreorderTreeIterator;
  39 import org.forester.util.ForesterUtil;
  40 import org.forester.util.ForesterUtil.PhylogenyNodeField;
  41
  42 public final class WebserviceUtil {
  43
  44     public static final String TAX_CODE_TO_SCI_NAME = "tax_code_to_sci_name";
  45     public static final String TREE_FAM_INST        = "tree_fam";
  46     public static final String PFAM_INST            = "pfam";
  47     public static final String TOL_WEBSERVER        = "http://tolweb.org/onlinecontributors/app?service=external&page=xml/TreeStructureService&node_id="
  48                                                             + PhylogeniesWebserviceClient.QUERY_PLACEHOLDER;
  49     public static final String TOL_NAME             = "Tree of Life";
  50     public static final String TREE_BASE_NAME       = "TreeBASE";
  51     public static final String TREE_FAM_NAME        = "TreeFam";
  52     public static final String PFAM_NAME            = "Pfam";
  53     public static final String PFAM_SERVER          = "http://pfam.janelia.org";
  54
  55     public static List<PhylogeniesWebserviceClient> createDefaultClients() {
  56         final List<PhylogeniesWebserviceClient> clients = new ArrayList<PhylogeniesWebserviceClient>();
  57         clients
  58                 .add( new BasicPhylogeniesWebserviceClient( TOL_NAME,
  59                                                             "Read Tree from Tree of Life...",
  60                                                             "Use ToL webservice to obtain a phylogeny",
  61                                                             "Please enter a Tree of Life node identifier\n(Examples: "
  62                                                                     + "19386 for Cephalopoda, 2461 for Cnidaria, 2466 for Deuterostomia)",
  63                                                             WsPhylogenyFormat.TOL_XML_RESPONSE,
  64                                                             PhylogenyNodeField.TAXONOMY_SCIENTIFIC_NAME,
  65                                                             WebserviceUtil.TOL_WEBSERVER,
  66                                                             true,
  67                                                             "http://tolweb.org",
  68                                                             null ) );
  69         clients
  70                 .add( new BasicPhylogeniesWebserviceClient( TREE_BASE_NAME,
  71                                                             "Read Tree from TreeBASE...",
  72                                                             "Use TreeBASE to obtain a phylogeny",
  73                                                             "Please enter a TreeBASE tree identifier\n(Examples: 2654, 825, 4931, 2518, 2406, 4934)",
  74                                                             WsPhylogenyFormat.NEXUS,
  75                                                             PhylogenyNodeField.TAXONOMY_SCIENTIFIC_NAME,
  76                                                             "http://purl.org/phylo/treebase/phylows/tree/TB2:Tr"
  77                                                                     + PhylogeniesWebserviceClient.QUERY_PLACEHOLDER
  78                                                                     + "?format=nexus",
  79                                                             true,
  80                                                             "http://treebase.nescent.org",
  81                                                             null ) );
  82         clients
  83                 .add( new BasicPhylogeniesWebserviceClient( PFAM_NAME,
  84                                                             "Read Gene Tree from Pfam...",
  85                                                             "Use  Pfam to obtain a (full) gene tree",
  86                                                             "Please enter a Pfam (PF) accession number\n(Examples: 01849 for NAC, 00452 for Bcl-2, 00046 for Homeobox)",
  87                                                             WsPhylogenyFormat.PFAM,
  88                                                             null,
  89                                                             PFAM_SERVER + "/family/tree/download?alnType=full&acc=PF"
  90                                                                     + PhylogeniesWebserviceClient.QUERY_PLACEHOLDER,
  91                                                             false,
  92                                                             PFAM_SERVER,
  93                                                             PFAM_INST ) );
  94         clients
  95                 .add( new BasicPhylogeniesWebserviceClient( TREE_FAM_NAME,
  96                                                             "Read Full Gene Tree from TreeFam...",
  97                                                             "Use TreeFam to obtain a (full) gene tree",
  98                                                             "Please enter a TreeFam (TF) accession number\n(Examples: 101004 for Cyclin D, 315938 for Hox, 105310 for Wnt)",
  99                                                             WsPhylogenyFormat.NHX,
 100                                                             null,
 101                                                             "http://www.treefam.org/cgi-bin/getdata.pl?ac=TF"
 102                                                                     + PhylogeniesWebserviceClient.QUERY_PLACEHOLDER
 103                                                                     + "&f=full.nhx",
 104                                                             true,
 105                                                             "http://www.treefam.org",
 106                                                             TREE_FAM_INST ) );
 107         clients
 108                 .add( new BasicPhylogeniesWebserviceClient( TREE_FAM_NAME,
 109                                                             "Read Clean Gene Tree from TreeFam...",
 110                                                             "Use TreeFam to obtain a (\"clean\") gene tree",
 111                                                             "Please enter a TreeFam (TF) accession number\n(Examples: 101004 for Cyclin D, 315938 for Hox, 105310 for Wnt)",
 112                                                             WsPhylogenyFormat.NHX,
 113                                                             null,
 114                                                             "http://www.treefam.org/cgi-bin/getdata.pl?ac=TF"
 115                                                                     + PhylogeniesWebserviceClient.QUERY_PLACEHOLDER
 116                                                                     + "&f=clean.nhx",
 117                                                             true,
 118                                                             "http://www.treefam.org",
 119                                                             TREE_FAM_INST ) );
 120         return clients;
 121     }
 122
 123     static void extractSpTremblAccFromNodeName( final Phylogeny phy, final String source ) {
 124         final PreorderTreeIterator it = new PreorderTreeIterator( phy );
 125         while ( it.hasNext() ) {
 126             final PhylogenyNode n = it.next();
 127             if ( !ForesterUtil.isEmpty( n.getName() ) ) {
 128                 final String name = n.getName();
 129                 final int i = name.lastIndexOf( "/" );
 130                 if ( i > 0 ) {
 131                     final String acc_str = name.substring( 0, i );
 132                     if ( !ForesterUtil.isEmpty( acc_str ) ) {
 133                         final Sequence seq = new Sequence();
 134                         final Accession acc = new Accession( acc_str, source );
 135                         seq.setAccession( acc );
 136                         n.getNodeData().setSequence( seq );
 137                     }
 138                 }
 139             }
 140         }
 141     }
 142
 143     public static void processInstructions( final PhylogeniesWebserviceClient client, final Phylogeny phylogeny ) {
 144         if ( client.getProcessingInstructions().equals( WebserviceUtil.TAX_CODE_TO_SCI_NAME ) ) {
 145             WebserviceUtil.transferTaxonomyCodeToScientificName( phylogeny );
 146         }
 147         else if ( client.getProcessingInstructions().equals( WebserviceUtil.TREE_FAM_INST ) ) {
 148             WebserviceUtil.transferInternalTaxonomyCodeToScientificName( phylogeny );
 149             WebserviceUtil.transferExternalScientificNameToTaxonomyCode( phylogeny );
 150             WebserviceUtil.transferSequenceNameToSequenceAccession( phylogeny, "ensembl" );
 151             WebserviceUtil.setTaxonomyIdentifierType( phylogeny, "ncbi" );
 152         }
 153         else if ( client.getProcessingInstructions().equals( WebserviceUtil.PFAM_INST ) ) {
 154             WebserviceUtil.extractSpTremblAccFromNodeName( phylogeny, "sptrembl" );
 155         }
 156     }
 157
 158     static void setTaxonomyIdentifierType( final Phylogeny phy, final String type ) {
 159         final PhylogenyNodeIterator it = phy.iteratorPostorder();
 160         while ( it.hasNext() ) {
 161             final PhylogenyNode n = it.next();
 162             if ( n.getNodeData().isHasTaxonomy() && ( n.getNodeData().getTaxonomy().getIdentifier() != null ) ) {
 163                 n.getNodeData().getTaxonomy().setIdentifier( new Identifier( n.getNodeData().getTaxonomy()
 164                         .getIdentifier().getValue(), type ) );
 165             }
 166         }
 167     }
 168
 169     static void transferExternalScientificNameToTaxonomyCode( final Phylogeny phy ) {
 170         final PhylogenyNodeIterator it = phy.iteratorPostorder();
 171         while ( it.hasNext() ) {
 172             final PhylogenyNode n = it.next();
 173             if ( n.isExternal() && n.getNodeData().isHasTaxonomy() ) {
 174                 final String name = n.getNodeData().getTaxonomy().getScientificName();
 175                 if ( !ForesterUtil.isEmpty( name ) && PhyloXmlUtil.TAXOMONY_CODE_PATTERN.matcher( name ).matches() ) {
 176                     n.getNodeData().getTaxonomy().setScientificName( "" );
 177                     n.getNodeData().getTaxonomy().setTaxonomyCode( name );
 178                 }
 179             }
 180         }
 181     }
 182
 183     static void transferInternalTaxonomyCodeToScientificName( final Phylogeny phy ) {
 184         final PhylogenyNodeIterator it = phy.iteratorPostorder();
 185         while ( it.hasNext() ) {
 186             final PhylogenyNode n = it.next();
 187             if ( !n.isExternal() && n.getNodeData().isHasTaxonomy() ) {
 188                 final String name = n.getNodeData().getTaxonomy().getTaxonomyCode();
 189                 if ( !ForesterUtil.isEmpty( name ) ) {
 190                     n.getNodeData().getTaxonomy().setScientificName( name );
 191                     n.getNodeData().getTaxonomy().setTaxonomyCode( "" );
 192                 }
 193             }
 194         }
 195     }
 196
 197     static void transferSequenceNameToSequenceAccession( final Phylogeny phy, final String source ) {
 198         final PhylogenyNodeIterator it = phy.iteratorPostorder();
 199         while ( it.hasNext() ) {
 200             final PhylogenyNode n = it.next();
 201             if ( n.getNodeData().isHasSequence() ) {
 202                 final String name = n.getNodeData().getSequence().getName();
 203                 if ( !ForesterUtil.isEmpty( name ) ) {
 204                     n.getNodeData().getSequence().setName( "" );
 205                     n.getNodeData().getSequence().setAccession( new Accession( name, source ) );
 206                 }
 207             }
 208         }
 209     }
 210
 211     static void transferTaxonomyCodeToScientificName( final Phylogeny phy ) {
 212         final PhylogenyNodeIterator it = phy.iteratorPostorder();
 213         while ( it.hasNext() ) {
 214             final PhylogenyNode n = it.next();
 215             if ( n.getNodeData().isHasTaxonomy() ) {
 216                 final String name = n.getNodeData().getTaxonomy().getTaxonomyCode();
 217                 if ( !ForesterUtil.isEmpty( name ) ) {
 218                     n.getNodeData().getTaxonomy().setScientificName( name );
 219                     n.getNodeData().getTaxonomy().setTaxonomyCode( "" );
 220                 }
 221             }
 222         }
 223     }
 224 }