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