git://source.jalview.org / jalview.git / blobdiff
? search:
summary | shortlog | log | commit | commitdiff | tree
raw | inline | side by side
(no commit message)
[jalview.git] / forester / java / src / org / forester / analysis / TaxonomyDataManager.java
diff --git a/forester/java/src/org/forester/analysis/TaxonomyDataManager.java b/forester/java/src/org/forester/analysis/TaxonomyDataManager.java
index 79ab713..4cc61cd 100644 (file)
--- a/forester/java/src/org/forester/analysis/TaxonomyDataManager.java
+++ b/forester/java/src/org/forester/analysis/TaxonomyDataManager.java
@@ -33,6 +33,7 @@ import java.util.HashMap;
 import java.util.List;
 import java.util.SortedSet;
 import java.util.TreeSet;
+import java.util.regex.Matcher;
 
 import javax.swing.JOptionPane;
 
@@ -40,13 +41,16 @@ import org.forester.archaeopteryx.MainFrameApplication;
 import org.forester.archaeopteryx.TreePanel;
 import org.forester.archaeopteryx.tools.AncestralTaxonomyInferrer;
 import org.forester.archaeopteryx.tools.RunnableProcess;
+import org.forester.io.parsers.nhx.NHXParser;
 import org.forester.io.parsers.phyloxml.PhyloXmlDataFormatException;
+import org.forester.io.parsers.util.ParserUtils;
 import org.forester.phylogeny.Phylogeny;
 import org.forester.phylogeny.PhylogenyNode;
 import org.forester.phylogeny.data.Identifier;
 import org.forester.phylogeny.data.Taxonomy;
 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
 import org.forester.util.ForesterUtil;
+import org.forester.util.TaxonomyUtil;
 import org.forester.ws.seqdb.SequenceDbWsTools;
 import org.forester.ws.seqdb.UniProtTaxonomy;
 
@@ -56,7 +60,7 @@ public final class TaxonomyDataManager extends RunnableProcess {
         CODE, SN, CN, ID, LIN;
     }
     private static final int                              MAX_CACHE_SIZE           = 100000;
-    private static final int                              MAX_TAXONOMIES_TO_RETURN = 10;
+    private static final int                              MAX_TAXONOMIES_TO_RETURN = 2000;
     private static final HashMap<String, UniProtTaxonomy> _sn_up_cache_map         = new HashMap<String, UniProtTaxonomy>();
     private static final HashMap<String, UniProtTaxonomy> _lineage_up_cache_map    = new HashMap<String, UniProtTaxonomy>();
     private static final HashMap<String, UniProtTaxonomy> _code_up_cache_map       = new HashMap<String, UniProtTaxonomy>();
@@ -129,7 +133,7 @@ public final class TaxonomyDataManager extends RunnableProcess {
     private final static UniProtTaxonomy obtainTaxonomy( final HashMap<String, UniProtTaxonomy> cache,
                                                          final Object query,
                                                          final QUERY_TYPE qt ) throws IOException,
-            AncestralTaxonomyInferenceException {
+                                                         AncestralTaxonomyInferenceException {
         if ( cache.containsKey( query ) ) {
             return cache.get( query ).copy();
         }
@@ -184,20 +188,24 @@ public final class TaxonomyDataManager extends RunnableProcess {
     }
 
     private final static List<UniProtTaxonomy> getTaxonomiesFromScientificName( final String query ) throws IOException {
-        if ( query.equalsIgnoreCase( UniProtTaxonomy.BACTERIA ) ||
-                query.equalsIgnoreCase( UniProtTaxonomy.ARCHAEA ) ||
-                query.equalsIgnoreCase( UniProtTaxonomy.VIRUSES ) ||
-                query.equalsIgnoreCase( UniProtTaxonomy.EUKARYOTA ) 
-                ) {
+        if ( query.equalsIgnoreCase( UniProtTaxonomy.BACTERIA ) || query.equalsIgnoreCase( UniProtTaxonomy.ARCHAEA )
+                || query.equalsIgnoreCase( UniProtTaxonomy.VIRUSES )
+                || query.equalsIgnoreCase( UniProtTaxonomy.EUKARYOTA ) || query.equalsIgnoreCase( UniProtTaxonomy.X ) ) {
             final List<UniProtTaxonomy> l = new ArrayList<UniProtTaxonomy>();
             l.add( UniProtTaxonomy.createSpecialFromScientificName( query ) );
             return l;
         }
-        
         return SequenceDbWsTools.getTaxonomiesFromScientificNameStrict( query, MAX_TAXONOMIES_TO_RETURN );
     }
 
     private final static List<UniProtTaxonomy> getTaxonomiesFromTaxonomyCode( final String query ) throws IOException {
+        //FIXME fix "SPHAR" issue
+        if ( ( ( query.indexOf( "XX" ) == 3 ) && TaxonomyUtil.isHasTaxIdFromFakeTaxCode( query ) )
+                || query.equals( "SPHAR" ) /* TODO remove me, is same as Sphingomonas aromaticivorans */
+                ) {
+            final int id = TaxonomyUtil.getTaxIdFromFakeTaxCode( query );
+            return SequenceDbWsTools.getTaxonomiesFromId( String.valueOf( id ), MAX_TAXONOMIES_TO_RETURN );
+        }
         return SequenceDbWsTools.getTaxonomiesFromTaxonomyCode( query, MAX_TAXONOMIES_TO_RETURN );
     }
 
@@ -211,7 +219,7 @@ public final class TaxonomyDataManager extends RunnableProcess {
     synchronized final private static SortedSet<String> obtainDetailedTaxonomicInformation( final Phylogeny phy,
                                                                                             final boolean delete,
                                                                                             final boolean allow_to_use_basic_node_names )
-            throws IOException, AncestralTaxonomyInferenceException {
+                                                                                                    throws IOException, AncestralTaxonomyInferenceException {
         clearCachesIfTooLarge();
         final SortedSet<String> not_found = new TreeSet<String>();
         List<PhylogenyNode> not_found_external_nodes = null;
@@ -243,7 +251,10 @@ public final class TaxonomyDataManager extends RunnableProcess {
             if ( ( ( tax != null ) && ( isHasAppropriateId( tax ) || !ForesterUtil.isEmpty( tax.getScientificName() )
                     || !ForesterUtil.isEmpty( tax.getTaxonomyCode() ) || !ForesterUtil.isEmpty( tax.getCommonName() ) ) )
                     || ( allow_to_use_basic_node_names && !ForesterUtil.isEmpty( node.getName() ) ) ) {
-                if ( tax != null ) {
+                if ( ( ( tax != null ) && ( isHasAppropriateId( tax )
+                        || !ForesterUtil.isEmpty( tax.getScientificName() )
+                        || !ForesterUtil.isEmpty( tax.getTaxonomyCode() ) || !ForesterUtil
+                        .isEmpty( tax.getCommonName() ) ) ) ) {
                     uniprot_tax = obtainUniProtTaxonomy( tax, null, qt );
                 }
                 else {
@@ -253,7 +264,6 @@ public final class TaxonomyDataManager extends RunnableProcess {
                     if ( tax == null ) {
                         tax = new Taxonomy();
                         node.getNodeData().addTaxonomy( tax );
-                        node.setName( "" );
                     }
                     updateTaxonomy( qt, node, tax, uniprot_tax );
                 }
@@ -320,15 +330,39 @@ public final class TaxonomyDataManager extends RunnableProcess {
         if ( ForesterUtil.isEmpty( simple_name ) ) {
             throw new IllegalArgumentException( "illegal attempt to use empty simple name" );
         }
-        qt = QUERY_TYPE.SN;
-        UniProtTaxonomy ut = obtainTaxonomy( TaxonomyDataManager.getSnTaxCacheMap(), simple_name, qt );
-        if ( ut == null ) {
+        UniProtTaxonomy ut = null;
+        final String code = ParserUtils.extractTaxonomyCodeFromNodeName( simple_name,
+                                                                         NHXParser.TAXONOMY_EXTRACTION.AGGRESSIVE );
+        if ( !ForesterUtil.isEmpty( code ) ) {
             qt = QUERY_TYPE.CODE;
-            ut = obtainTaxonomy( TaxonomyDataManager.getCodeTaxCacheMap(), simple_name, qt );
+            ut = obtainTaxonomy( TaxonomyDataManager.getCodeTaxCacheMap(), code, qt );
         }
         if ( ut == null ) {
-            qt = QUERY_TYPE.CN;
-            ut = obtainTaxonomy( TaxonomyDataManager.getCnTaxCacheMap(), simple_name, qt );
+            final String sn = ParserUtils.extractScientificNameFromNodeName( simple_name );
+            if ( !ForesterUtil.isEmpty( sn ) ) {
+                qt = QUERY_TYPE.SN;
+                ut = obtainTaxonomy( TaxonomyDataManager.getSnTaxCacheMap(), sn, qt );
+            }
+        }
+        if ( ut == null ) {
+            final String id = ParserUtils
+                    .extractUniprotTaxonomyIdFromNodeName( simple_name,
+                                                           NHXParser.TAXONOMY_EXTRACTION.PFAM_STYLE_RELAXED );
+            if ( !ForesterUtil.isEmpty( id ) ) {
+                qt = QUERY_TYPE.ID;
+                ut = obtainTaxonomy( TaxonomyDataManager.getIdTaxCacheMap(), id, qt );
+            }
+        }
+        if ( ut == null ) {
+            String sn = "";
+            final Matcher m = ParserUtils.TAXOMONY_SN_PATTERN_GENUS.matcher( simple_name );
+            if ( m.matches() ) {
+                sn = m.group( 1 );
+            }
+            if ( !ForesterUtil.isEmpty( sn ) ) {
+                qt = QUERY_TYPE.SN;
+                ut = obtainTaxonomy( TaxonomyDataManager.getSnTaxCacheMap(), sn, qt );
+            }
         }
         return ut;
     }
@@ -342,7 +376,7 @@ public final class TaxonomyDataManager extends RunnableProcess {
         else {
             final List<UniProtTaxonomy> matching_taxonomies = new ArrayList<UniProtTaxonomy>();
             final List<UniProtTaxonomy> up_taxonomies = getTaxonomiesFromScientificName( lineage
-                    .get( lineage.size() - 1 ) );
+                                                                                         .get( lineage.size() - 1 ) );
             if ( ( up_taxonomies != null ) && ( up_taxonomies.size() > 0 ) ) {
                 for( final UniProtTaxonomy up_taxonomy : up_taxonomies ) {
                     boolean match = true;
@@ -391,7 +425,7 @@ public final class TaxonomyDataManager extends RunnableProcess {
             }
             else {
                 throw new AncestralTaxonomyInferenceException( "taxonomy \"" + ( lineage.get( lineage.size() - 1 ) )
-                        + "\" not found" );
+                                                               + "\" not found" );
             }
         }
     }
@@ -400,7 +434,7 @@ public final class TaxonomyDataManager extends RunnableProcess {
                                                            final PhylogenyNode node,
                                                            final Taxonomy tax,
                                                            final UniProtTaxonomy up_tax )
-            throws PhyloXmlDataFormatException {
+                                                                   throws PhyloXmlDataFormatException {
         if ( ( qt != QUERY_TYPE.SN ) && !ForesterUtil.isEmpty( up_tax.getScientificName() )
                 && ForesterUtil.isEmpty( tax.getScientificName() ) ) {
             tax.setScientificName( up_tax.getScientificName() );
@@ -478,7 +512,7 @@ public final class TaxonomyDataManager extends RunnableProcess {
                                                JOptionPane.WARNING_MESSAGE );
             }
             catch ( final Exception e ) {
-                // Not important if this fails, do nothing. 
+                // Not important if this fails, do nothing.
             }
             return;
         }
@@ -530,7 +564,7 @@ public final class TaxonomyDataManager extends RunnableProcess {
                                                JOptionPane.WARNING_MESSAGE );
             }
             catch ( final Exception e ) {
-                // Not important if this fails, do nothing. 
+                // Not important if this fails, do nothing.
             }
         }
         else {