X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=forester%2Fjava%2Fsrc%2Forg%2Fforester%2Ftest%2FTest.java;h=64b7807f23b2fedee2f406f6ea2bfc146557c6ee;hb=482c9a54ff2b83c947c43449c7b0d86dc9c8dafd;hp=796acd06d035a8f4d9b5dd1caeb0c5af6106bf01;hpb=7bbb45295594e9574ee64a32fac17a93b5aa6ade;p=jalview.git

diff --git a/forester/java/src/org/forester/test/Test.java b/forester/java/src/org/forester/test/Test.java
index 796acd0..64b7807 100644
--- a/forester/java/src/org/forester/test/Test.java
+++ b/forester/java/src/org/forester/test/Test.java
@@ -36,6 +36,7 @@ import java.util.Iterator;
 import java.util.List;
 import java.util.Locale;
 import java.util.Set;
+import java.util.SortedSet;
 
 import org.forester.application.support_transfer;
 import org.forester.archaeopteryx.TreePanelUtil;
@@ -70,6 +71,7 @@ import org.forester.phylogeny.PhylogenyMethods;
 import org.forester.phylogeny.PhylogenyNode;
 import org.forester.phylogeny.PhylogenyNode.NH_CONVERSION_SUPPORT_VALUE_STYLE;
 import org.forester.phylogeny.data.Accession;
+import org.forester.phylogeny.data.Accession.Source;
 import org.forester.phylogeny.data.BinaryCharacters;
 import org.forester.phylogeny.data.BranchWidth;
 import org.forester.phylogeny.data.Confidence;
@@ -113,7 +115,7 @@ import org.forester.util.DescriptiveStatistics;
 import org.forester.util.ForesterConstants;
 import org.forester.util.ForesterUtil;
 import org.forester.util.GeneralTable;
-import org.forester.util.SequenceIdParser;
+import org.forester.util.SequenceAccessionTools;
 import org.forester.ws.seqdb.SequenceDatabaseEntry;
 import org.forester.ws.seqdb.SequenceDbWsTools;
 import org.forester.ws.seqdb.UniProtTaxonomy;
@@ -125,6 +127,7 @@ import org.forester.ws.wabi.TxSearch.TAX_RANK;
 @SuppressWarnings( "unused")
 public final class Test {
 
+    private final static boolean PERFORM_DB_TESTS          = true;
     private final static double  ZERO_DIFF                 = 1.0E-9;
     private final static String  PATH_TO_TEST_DATA         = System.getProperty( "user.dir" )
                                                                    + ForesterUtil.getFileSeparator() + "test_data"
@@ -171,6 +174,15 @@ public final class Test {
             System.exit( -1 );
         }
         final long start_time = new Date().getTime();
+        System.out.print( "Basic node methods: " );
+        if ( Test.testBasicNodeMethods() ) {
+            System.out.println( "OK." );
+            succeeded++;
+        }
+        else {
+            System.out.println( "failed." );
+            failed++;
+        }
         System.out.print( "Protein id: " );
         if ( !testProteinId() ) {
             System.out.println( "failed." );
@@ -225,8 +237,8 @@ public final class Test {
             System.out.println( "failed." );
             failed++;
         }
-        System.out.print( "Hmmscan output parser: " );
-        if ( testHmmscanOutputParser() ) {
+        System.out.print( "UniProtKB id extraction: " );
+        if ( Test.testExtractUniProtKbProteinSeqIdentifier() ) {
             System.out.println( "OK." );
             succeeded++;
         }
@@ -234,8 +246,8 @@ public final class Test {
             System.out.println( "failed." );
             failed++;
         }
-        System.out.print( "Basic node methods: " );
-        if ( Test.testBasicNodeMethods() ) {
+        System.out.print( "Sequence DB tools 1: " );
+        if ( testSequenceDbWsTools1() ) {
             System.out.println( "OK." );
             succeeded++;
         }
@@ -243,8 +255,21 @@ public final class Test {
             System.out.println( "failed." );
             failed++;
         }
-        System.out.print( "Taxonomy code extraction: " );
-        if ( Test.testExtractTaxonomyCodeFromNodeName() ) {
+        if ( PERFORM_DB_TESTS ) {
+            System.out.print( "Sequence DB tools 2: " );
+            if ( testSequenceDbWsTools2() ) {
+                System.out.println( "OK." );
+                succeeded++;
+            }
+            else {
+                System.out.println( "failed." );
+                failed++;
+                System.exit( -1 );
+            }
+        }
+        System.exit( 0 );
+        System.out.print( "Hmmscan output parser: " );
+        if ( testHmmscanOutputParser() ) {
             System.out.println( "OK." );
             succeeded++;
         }
@@ -252,8 +277,8 @@ public final class Test {
             System.out.println( "failed." );
             failed++;
         }
-        System.out.print( "SN extraction: " );
-        if ( Test.testExtractSNFromNodeName() ) {
+        System.out.print( "Taxonomy code extraction: " );
+        if ( Test.testExtractTaxonomyCodeFromNodeName() ) {
             System.out.println( "OK." );
             succeeded++;
         }
@@ -261,8 +286,8 @@ public final class Test {
             System.out.println( "failed." );
             failed++;
         }
-        System.out.print( "Taxonomy extraction (general): " );
-        if ( Test.testTaxonomyExtraction() ) {
+        System.out.print( "SN extraction: " );
+        if ( Test.testExtractSNFromNodeName() ) {
             System.out.println( "OK." );
             succeeded++;
         }
@@ -270,8 +295,8 @@ public final class Test {
             System.out.println( "failed." );
             failed++;
         }
-        System.out.print( "UniProtKB id extraction: " );
-        if ( Test.testExtractUniProtKbProteinSeqIdentifier() ) {
+        System.out.print( "Taxonomy extraction (general): " );
+        if ( Test.testTaxonomyExtraction() ) {
             System.out.println( "OK." );
             succeeded++;
         }
@@ -813,23 +838,27 @@ public final class Test {
             System.out.println( "failed." );
             failed++;
         }
-        System.out.print( "Uniprot Entry Retrieval: " );
-        if ( Test.testUniprotEntryRetrieval() ) {
-            System.out.println( "OK." );
-            succeeded++;
-        }
-        else {
-            System.out.println( "failed." );
-            failed++;
-        }
-        System.out.print( "Uniprot Taxonomy Search: " );
-        if ( Test.testUniprotTaxonomySearch() ) {
-            System.out.println( "OK." );
-            succeeded++;
+        if ( PERFORM_DB_TESTS ) {
+            System.out.print( "Uniprot Entry Retrieval: " );
+            if ( Test.testUniprotEntryRetrieval() ) {
+                System.out.println( "OK." );
+                succeeded++;
+            }
+            else {
+                System.out.println( "failed." );
+                failed++;
+            }
         }
-        else {
-            System.out.println( "failed." );
-            failed++;
+        if ( PERFORM_DB_TESTS ) {
+            System.out.print( "Uniprot Taxonomy Search: " );
+            if ( Test.testUniprotTaxonomySearch() ) {
+                System.out.println( "OK." );
+                succeeded++;
+            }
+            else {
+                System.out.println( "failed." );
+                failed++;
+            }
         }
         //----
         String path = "";
@@ -1208,6 +1237,22 @@ public final class Test {
             if ( !( t3.getNode( "root node" ).getNodeData().getSequence().getLocation().equals( "12p13-p12" ) ) ) {
                 return false;
             }
+            final SortedSet<Accession> x = t3.getNode( "root node" ).getNodeData().getSequence().getCrossReferences();
+            if ( x.size() != 4 ) {
+                return false;
+            }
+            int c = 0;
+            for( final Accession acc : x ) {
+                if ( c == 0 ) {
+                    if ( !acc.getSource().equals( "KEGG" ) ) {
+                        return false;
+                    }
+                    if ( !acc.getValue().equals( "hsa:596" ) ) {
+                        return false;
+                    }
+                }
+                c++;
+            }
         }
         catch ( final Exception e ) {
             e.printStackTrace( System.out );
@@ -1487,7 +1532,6 @@ public final class Test {
             }
             if ( ( ( BinaryCharacters ) t3_rt.getNode( "node bb" ).getNodeData().getBinaryCharacters().copy() )
                     .getLostCount() != BinaryCharacters.COUNT_DEFAULT ) {
-                ;
                 return false;
             }
             if ( t3_rt.getNode( "node b" ).getNodeData().getBinaryCharacters().getGainedCount() != 1 ) {
@@ -1536,6 +1580,23 @@ public final class Test {
                     .equalsIgnoreCase( "433" ) ) {
                 return false;
             }
+            final SortedSet<Accession> x = t3_rt.getNode( "root node" ).getNodeData().getSequence()
+                    .getCrossReferences();
+            if ( x.size() != 4 ) {
+                return false;
+            }
+            int c = 0;
+            for( final Accession acc : x ) {
+                if ( c == 0 ) {
+                    if ( !acc.getSource().equals( "KEGG" ) ) {
+                        return false;
+                    }
+                    if ( !acc.getValue().equals( "hsa:596" ) ) {
+                        return false;
+                    }
+                }
+                c++;
+            }
         }
         catch ( final Exception e ) {
             e.printStackTrace( System.out );
@@ -2471,8 +2532,8 @@ public final class Test {
             if ( !TreePanelUtil.createUriForSeqWeb( n, null, null ).equals( ForesterUtil.NCBI_PROTEIN + "AAA34956" ) ) {
                 return false;
             }
-            n.setName( "j40f4_Q06891.1_fndn2 fnr3" );
-            if ( !TreePanelUtil.createUriForSeqWeb( n, null, null ).equals( ForesterUtil.NCBI_PROTEIN + "Q06891.1" ) ) {
+            n.setName( "AAA34956" );
+            if ( !TreePanelUtil.createUriForSeqWeb( n, null, null ).equals( ForesterUtil.NCBI_PROTEIN + "AAA34956" ) ) {
                 return false;
             }
             n.setName( "GI:394892" );
@@ -2490,6 +2551,16 @@ public final class Test {
                 System.out.println( TreePanelUtil.createUriForSeqWeb( n, null, null ) );
                 return false;
             }
+            n.setName( "P12345" );
+            if ( !TreePanelUtil.createUriForSeqWeb( n, null, null ).equals( ForesterUtil.UNIPROT_KB + "P12345" ) ) {
+                System.out.println( TreePanelUtil.createUriForSeqWeb( n, null, null ) );
+                return false;
+            }
+            n.setName( "gi_fdgjmn-3jk5-243 mnefmn fg023-0 P12345 4395jtmnsrg02345m1ggi92450jrg890j4t0j240" );
+            if ( !TreePanelUtil.createUriForSeqWeb( n, null, null ).equals( ForesterUtil.UNIPROT_KB + "P12345" ) ) {
+                System.out.println( TreePanelUtil.createUriForSeqWeb( n, null, null ) );
+                return false;
+            }
         }
         catch ( final Exception e ) {
             e.printStackTrace( System.out );
@@ -3296,46 +3367,46 @@ public final class Test {
         //Nucleotide: 1 letter + 5 numerals OR 2 letters + 6 numerals
         //Protein:    3 letters + 5 numerals
         //http://www.ncbi.nlm.nih.gov/Sequin/acc.html
-        if ( !SequenceIdParser.parseGenbankAccessor( "AY423861" ).equals( "AY423861" ) ) {
+        if ( !SequenceAccessionTools.parseGenbankAccessorFromString( "AY423861" ).equals( "AY423861" ) ) {
             return false;
         }
-        if ( !SequenceIdParser.parseGenbankAccessor( ".AY423861.2" ).equals( "AY423861.2" ) ) {
+        if ( !SequenceAccessionTools.parseGenbankAccessorFromString( ".AY423861.2" ).equals( "AY423861.2" ) ) {
             return false;
         }
-        if ( !SequenceIdParser.parseGenbankAccessor( "345_.AY423861.24_345" ).equals( "AY423861.24" ) ) {
+        if ( !SequenceAccessionTools.parseGenbankAccessorFromString( "345_.AY423861.24_345" ).equals( "AY423861.24" ) ) {
             return false;
         }
-        if ( SequenceIdParser.parseGenbankAccessor( "AAY423861" ) != null ) {
+        if ( SequenceAccessionTools.parseGenbankAccessorFromString( "AAY423861" ) != null ) {
             return false;
         }
-        if ( SequenceIdParser.parseGenbankAccessor( "AY4238612" ) != null ) {
+        if ( SequenceAccessionTools.parseGenbankAccessorFromString( "AY4238612" ) != null ) {
             return false;
         }
-        if ( SequenceIdParser.parseGenbankAccessor( "AAY4238612" ) != null ) {
+        if ( SequenceAccessionTools.parseGenbankAccessorFromString( "AAY4238612" ) != null ) {
             return false;
         }
-        if ( SequenceIdParser.parseGenbankAccessor( "Y423861" ) != null ) {
+        if ( SequenceAccessionTools.parseGenbankAccessorFromString( "Y423861" ) != null ) {
             return false;
         }
-        if ( !SequenceIdParser.parseGenbankAccessor( "S12345" ).equals( "S12345" ) ) {
+        if ( !SequenceAccessionTools.parseGenbankAccessorFromString( "S12345" ).equals( "S12345" ) ) {
             return false;
         }
-        if ( !SequenceIdParser.parseGenbankAccessor( "|S12345|" ).equals( "S12345" ) ) {
+        if ( !SequenceAccessionTools.parseGenbankAccessorFromString( "|S12345|" ).equals( "S12345" ) ) {
             return false;
         }
-        if ( SequenceIdParser.parseGenbankAccessor( "|S123456" ) != null ) {
+        if ( SequenceAccessionTools.parseGenbankAccessorFromString( "|S123456" ) != null ) {
             return false;
         }
-        if ( SequenceIdParser.parseGenbankAccessor( "ABC123456" ) != null ) {
+        if ( SequenceAccessionTools.parseGenbankAccessorFromString( "ABC123456" ) != null ) {
             return false;
         }
-        if ( !SequenceIdParser.parseGenbankAccessor( "ABC12345" ).equals( "ABC12345" ) ) {
+        if ( !SequenceAccessionTools.parseGenbankAccessorFromString( "ABC12345" ).equals( "ABC12345" ) ) {
             return false;
         }
-        if ( !SequenceIdParser.parseGenbankAccessor( "&ABC12345&" ).equals( "ABC12345" ) ) {
+        if ( !SequenceAccessionTools.parseGenbankAccessorFromString( "&ABC12345&" ).equals( "ABC12345" ) ) {
             return false;
         }
-        if ( SequenceIdParser.parseGenbankAccessor( "ABCD12345" ) != null ) {
+        if ( SequenceAccessionTools.parseGenbankAccessorFromString( "ABCD12345" ) != null ) {
             return false;
         }
         return true;
@@ -3629,166 +3700,166 @@ public final class Test {
         try {
             PhylogenyNode n = new PhylogenyNode();
             n.setName( "tr|B3RJ64" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "B3RJ64" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "B3RJ64" ) ) {
                 return false;
             }
             n.setName( "tr.B3RJ64" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "B3RJ64" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "B3RJ64" ) ) {
                 return false;
             }
             n.setName( "tr=B3RJ64" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "B3RJ64" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "B3RJ64" ) ) {
                 return false;
             }
             n.setName( "tr-B3RJ64" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "B3RJ64" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "B3RJ64" ) ) {
                 return false;
             }
             n.setName( "tr/B3RJ64" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "B3RJ64" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "B3RJ64" ) ) {
                 return false;
             }
             n.setName( "tr\\B3RJ64" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "B3RJ64" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "B3RJ64" ) ) {
                 return false;
             }
             n.setName( "tr_B3RJ64" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "B3RJ64" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "B3RJ64" ) ) {
                 return false;
             }
             n.setName( " tr|B3RJ64 " );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "B3RJ64" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "B3RJ64" ) ) {
                 return false;
             }
             n.setName( "-tr|B3RJ64-" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "B3RJ64" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "B3RJ64" ) ) {
                 return false;
             }
             n.setName( "-tr=B3RJ64-" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "B3RJ64" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "B3RJ64" ) ) {
                 return false;
             }
             n.setName( "_tr=B3RJ64_" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "B3RJ64" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "B3RJ64" ) ) {
                 return false;
             }
             n.setName( " tr_tr|B3RJ64_sp|123 " );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "B3RJ64" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "B3RJ64" ) ) {
                 return false;
             }
-            n.setName( "sp|B3RJ64" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "B3RJ64" ) ) {
+            n.setName( "B3RJ64" );
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "B3RJ64" ) ) {
                 return false;
             }
-            n.setName( "ssp|B3RJ64" );
-            if ( ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ) != null ) {
+            n.setName( "sp|B3RJ64" );
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "B3RJ64" ) ) {
                 return false;
             }
             n.setName( "sp|B3RJ64C" );
-            if ( ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ) != null ) {
+            if ( SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ) != null ) {
                 return false;
             }
             n.setName( "sp B3RJ64" );
-            if ( ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ) != null ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "B3RJ64" ) ) {
                 return false;
             }
             n.setName( "sp|B3RJ6X" );
-            if ( ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ) != null ) {
+            if ( SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ) != null ) {
                 return false;
             }
             n.setName( "sp|B3RJ6" );
-            if ( ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ) != null ) {
+            if ( SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ) != null ) {
                 return false;
             }
             n.setName( "K1PYK7_CRAGI" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "K1PYK7_CRAGI" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "K1PYK7_CRAGI" ) ) {
                 return false;
             }
             n.setName( "K1PYK7_PEA" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "K1PYK7_PEA" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "K1PYK7_PEA" ) ) {
                 return false;
             }
             n.setName( "K1PYK7_RAT" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "K1PYK7_RAT" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "K1PYK7_RAT" ) ) {
                 return false;
             }
             n.setName( "K1PYK7_PIG" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "K1PYK7_PIG" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "K1PYK7_PIG" ) ) {
                 return false;
             }
             n.setName( "~K1PYK7_PIG~" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "K1PYK7_PIG" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "K1PYK7_PIG" ) ) {
                 return false;
             }
             n.setName( "123456_ECOLI-K1PYK7_CRAGI-sp" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "K1PYK7_CRAGI" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "K1PYK7_CRAGI" ) ) {
                 return false;
             }
             n.setName( "K1PYKX_CRAGI" );
-            if ( ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ) != null ) {
+            if ( SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ) != null ) {
                 return false;
             }
             n.setName( "XXXXX_CRAGI" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "XXXXX_CRAGI" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "XXXXX_CRAGI" ) ) {
                 return false;
             }
             n.setName( "tr|H3IB65|H3IB65_STRPU~2-2" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "H3IB65" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "H3IB65" ) ) {
                 return false;
             }
             n.setName( "jgi|Lacbi2|181470|Lacbi1.estExt_GeneWisePlus_human.C_10729~2-3" );
-            if ( ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ) != null ) {
+            if ( SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ) != null ) {
                 return false;
             }
             n.setName( "sp|Q86U06|RBM23_HUMAN~2-2" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "Q86U06" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "Q86U06" ) ) {
                 return false;
             }
             n = new PhylogenyNode();
             org.forester.phylogeny.data.Sequence seq = new org.forester.phylogeny.data.Sequence();
             seq.setSymbol( "K1PYK7_CRAGI" );
             n.getNodeData().addSequence( seq );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "K1PYK7_CRAGI" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "K1PYK7_CRAGI" ) ) {
                 return false;
             }
             seq.setSymbol( "tr|B3RJ64" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "B3RJ64" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "B3RJ64" ) ) {
                 return false;
             }
             n = new PhylogenyNode();
             seq = new org.forester.phylogeny.data.Sequence();
             seq.setName( "K1PYK7_CRAGI" );
             n.getNodeData().addSequence( seq );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "K1PYK7_CRAGI" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "K1PYK7_CRAGI" ) ) {
                 return false;
             }
             seq.setName( "tr|B3RJ64" );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "B3RJ64" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "B3RJ64" ) ) {
                 return false;
             }
             n = new PhylogenyNode();
             seq = new org.forester.phylogeny.data.Sequence();
             seq.setAccession( new Accession( "K1PYK8_CRAGI", "?" ) );
             n.getNodeData().addSequence( seq );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "K1PYK8_CRAGI" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "K1PYK8_CRAGI" ) ) {
                 return false;
             }
             n = new PhylogenyNode();
             seq = new org.forester.phylogeny.data.Sequence();
             seq.setAccession( new Accession( "tr|B3RJ64", "?" ) );
             n.getNodeData().addSequence( seq );
-            if ( !ForesterUtil.extractUniProtKbProteinSeqIdentifier( n ).equals( "B3RJ64" ) ) {
+            if ( !SequenceAccessionTools.obtainUniProtAccessorFromDataFields( n ).equals( "B3RJ64" ) ) {
                 return false;
             }
             //
             n = new PhylogenyNode();
             n.setName( "ACP19736" );
-            if ( !ForesterUtil.extractGenbankAccessor( n ).equals( "ACP19736" ) ) {
+            if ( !SequenceAccessionTools.obtainGenbankAccessorFromDataFields( n ).equals( "ACP19736" ) ) {
                 return false;
             }
             n = new PhylogenyNode();
-            n.setName( "_ACP19736_" );
-            if ( !ForesterUtil.extractGenbankAccessor( n ).equals( "ACP19736" ) ) {
+            n.setName( "|ACP19736|" );
+            if ( !SequenceAccessionTools.obtainGenbankAccessorFromDataFields( n ).equals( "ACP19736" ) ) {
                 return false;
             }
         }
@@ -9587,123 +9658,111 @@ public final class Test {
 
     private static boolean testSequenceIdParsing() {
         try {
-            Identifier id = SequenceIdParser.parse( "gb_ADF31344_segmented_worms_" );
-            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getProvider() )
-                    || !id.getValue().equals( "ADF31344" ) || !id.getProvider().equals( "ncbi" ) ) {
+            Accession id = SequenceAccessionTools.parseAccessorFromString( "gb_ADF31344_segmented_worms_" );
+            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getSource() )
+                    || !id.getValue().equals( "ADF31344" ) || !id.getSource().equals( "ncbi" ) ) {
                 if ( id != null ) {
                     System.out.println( "value   =" + id.getValue() );
-                    System.out.println( "provider=" + id.getProvider() );
+                    System.out.println( "provider=" + id.getSource() );
                 }
                 return false;
             }
             //
-            id = SequenceIdParser.parse( "segmented worms|gb_ADF31344" );
-            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getProvider() )
-                    || !id.getValue().equals( "ADF31344" ) || !id.getProvider().equals( "ncbi" ) ) {
+            id = SequenceAccessionTools.parseAccessorFromString( "segmented worms|gb_ADF31344" );
+            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getSource() )
+                    || !id.getValue().equals( "ADF31344" ) || !id.getSource().equals( "ncbi" ) ) {
                 if ( id != null ) {
                     System.out.println( "value   =" + id.getValue() );
-                    System.out.println( "provider=" + id.getProvider() );
+                    System.out.println( "provider=" + id.getSource() );
                 }
                 return false;
             }
             //
-            id = SequenceIdParser.parse( "segmented worms gb_ADF31344 and more" );
-            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getProvider() )
-                    || !id.getValue().equals( "ADF31344" ) || !id.getProvider().equals( "ncbi" ) ) {
-                if ( id != null ) {
-                    System.out.println( "value   =" + id.getValue() );
-                    System.out.println( "provider=" + id.getProvider() );
-                }
-                return false;
-            }
-            // 
-            id = SequenceIdParser.parse( "gb_AAA96518_1" );
-            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getProvider() )
-                    || !id.getValue().equals( "AAA96518" ) || !id.getProvider().equals( "ncbi" ) ) {
+            id = SequenceAccessionTools.parseAccessorFromString( "segmented worms gb_ADF31344 and more" );
+            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getSource() )
+                    || !id.getValue().equals( "ADF31344" ) || !id.getSource().equals( "ncbi" ) ) {
                 if ( id != null ) {
                     System.out.println( "value   =" + id.getValue() );
-                    System.out.println( "provider=" + id.getProvider() );
+                    System.out.println( "provider=" + id.getSource() );
                 }
                 return false;
             }
             // 
-            id = SequenceIdParser.parse( "gb_EHB07727_1_rodents_" );
-            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getProvider() )
-                    || !id.getValue().equals( "EHB07727" ) || !id.getProvider().equals( "ncbi" ) ) {
+            id = SequenceAccessionTools.parseAccessorFromString( "gb_AAA96518_1" );
+            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getSource() )
+                    || !id.getValue().equals( "AAA96518" ) || !id.getSource().equals( "ncbi" ) ) {
                 if ( id != null ) {
                     System.out.println( "value   =" + id.getValue() );
-                    System.out.println( "provider=" + id.getProvider() );
+                    System.out.println( "provider=" + id.getSource() );
                 }
                 return false;
             }
             // 
-            id = SequenceIdParser.parse( "dbj_BAF37827_1_turtles_" );
-            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getProvider() )
-                    || !id.getValue().equals( "BAF37827" ) || !id.getProvider().equals( "ncbi" ) ) {
+            id = SequenceAccessionTools.parseAccessorFromString( "gb_EHB07727_1_rodents_" );
+            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getSource() )
+                    || !id.getValue().equals( "EHB07727" ) || !id.getSource().equals( "ncbi" ) ) {
                 if ( id != null ) {
                     System.out.println( "value   =" + id.getValue() );
-                    System.out.println( "provider=" + id.getProvider() );
+                    System.out.println( "provider=" + id.getSource() );
                 }
                 return false;
             }
             // 
-            id = SequenceIdParser.parse( "emb_CAA73223_1_primates_" );
-            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getProvider() )
-                    || !id.getValue().equals( "CAA73223" ) || !id.getProvider().equals( "ncbi" ) ) {
+            id = SequenceAccessionTools.parseAccessorFromString( "dbj_BAF37827_1_turtles_" );
+            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getSource() )
+                    || !id.getValue().equals( "BAF37827" ) || !id.getSource().equals( "ncbi" ) ) {
                 if ( id != null ) {
                     System.out.println( "value   =" + id.getValue() );
-                    System.out.println( "provider=" + id.getProvider() );
+                    System.out.println( "provider=" + id.getSource() );
                 }
                 return false;
             }
             // 
-            id = SequenceIdParser.parse( "mites|ref_XP_002434188_1" );
-            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getProvider() )
-                    || !id.getValue().equals( "XP_002434188" ) || !id.getProvider().equals( "refseq" ) ) {
+            id = SequenceAccessionTools.parseAccessorFromString( "emb_CAA73223_1_primates_" );
+            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getSource() )
+                    || !id.getValue().equals( "CAA73223" ) || !id.getSource().equals( "ncbi" ) ) {
                 if ( id != null ) {
                     System.out.println( "value   =" + id.getValue() );
-                    System.out.println( "provider=" + id.getProvider() );
+                    System.out.println( "provider=" + id.getSource() );
                 }
                 return false;
             }
             // 
-            id = SequenceIdParser.parse( "mites_ref_XP_002434188_1_bla_XP_12345" );
-            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getProvider() )
-                    || !id.getValue().equals( "XP_002434188" ) || !id.getProvider().equals( "refseq" ) ) {
+            id = SequenceAccessionTools.parseAccessorFromString( "mites|ref_XP_002434188_1" );
+            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getSource() )
+                    || !id.getValue().equals( "XP_002434188" ) || !id.getSource().equals( "refseq" ) ) {
                 if ( id != null ) {
                     System.out.println( "value   =" + id.getValue() );
-                    System.out.println( "provider=" + id.getProvider() );
+                    System.out.println( "provider=" + id.getSource() );
                 }
                 return false;
             }
             // 
-            id = SequenceIdParser.parse( "P4A123" );
-            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getProvider() )
-                    || !id.getValue().equals( "P4A123" ) || !id.getProvider().equals( "sp" ) ) {
+            id = SequenceAccessionTools.parseAccessorFromString( "mites_ref_XP_002434188_1_bla_XP_12345" );
+            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getSource() )
+                    || !id.getValue().equals( "XP_002434188" ) || !id.getSource().equals( "refseq" ) ) {
                 if ( id != null ) {
                     System.out.println( "value   =" + id.getValue() );
-                    System.out.println( "provider=" + id.getProvider() );
+                    System.out.println( "provider=" + id.getSource() );
                 }
                 return false;
             }
             // 
-            id = SequenceIdParser.parse( "pllf[pok P4A123_osdjfosnqo035-9233332904i000490 vf tmv x45" );
-            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getProvider() )
-                    || !id.getValue().equals( "P4A123" ) || !id.getProvider().equals( "sp" ) ) {
+            id = SequenceAccessionTools.parseAccessorFromString( "P4A123" );
+            if ( ( id == null ) || ForesterUtil.isEmpty( id.getValue() ) || ForesterUtil.isEmpty( id.getSource() )
+                    || !id.getValue().equals( "P4A123" ) || !id.getSource().equals( "uniprot" ) ) {
                 if ( id != null ) {
                     System.out.println( "value   =" + id.getValue() );
-                    System.out.println( "provider=" + id.getProvider() );
+                    System.out.println( "provider=" + id.getSource() );
                 }
                 return false;
             }
-            // 
-            id = SequenceIdParser.parse( "XP_12345" );
+            id = SequenceAccessionTools.parseAccessorFromString( "XP_12345" );
             if ( id != null ) {
                 System.out.println( "value   =" + id.getValue() );
-                System.out.println( "provider=" + id.getProvider() );
+                System.out.println( "provider=" + id.getSource() );
                 return false;
             }
-            // lcl_91970_unknown_
         }
         catch ( final Exception e ) {
             e.printStackTrace( System.out );
@@ -10806,6 +10865,179 @@ public final class Test {
         return true;
     }
 
+    private static boolean testSequenceDbWsTools1() {
+        try {
+            final PhylogenyNode n = new PhylogenyNode();
+            n.setName( "NP_001025424" );
+            Accession acc = SequenceDbWsTools.obtainSeqAccession( n );
+            if ( ( acc == null ) || !acc.getSource().equals( Source.REFSEQ.toString() )
+                    || !acc.getValue().equals( "NP_001025424" ) ) {
+                return false;
+            }
+            n.setName( "340 0559 -- _NP_001025424_dsfdg15 05" );
+            acc = SequenceDbWsTools.obtainSeqAccession( n );
+            if ( ( acc == null ) || !acc.getSource().equals( Source.REFSEQ.toString() )
+                    || !acc.getValue().equals( "NP_001025424" ) ) {
+                return false;
+            }
+            n.setName( "NP_001025424.1" );
+            acc = SequenceDbWsTools.obtainSeqAccession( n );
+            if ( ( acc == null ) || !acc.getSource().equals( Source.REFSEQ.toString() )
+                    || !acc.getValue().equals( "NP_001025424" ) ) {
+                return false;
+            }
+            n.setName( "NM_001030253" );
+            acc = SequenceDbWsTools.obtainSeqAccession( n );
+            if ( ( acc == null ) || !acc.getSource().equals( Source.REFSEQ.toString() )
+                    || !acc.getValue().equals( "NM_001030253" ) ) {
+                return false;
+            }
+            n.setName( "BCL2_HUMAN" );
+            acc = SequenceDbWsTools.obtainSeqAccession( n );
+            if ( ( acc == null ) || !acc.getSource().equals( Source.UNIPROT.toString() )
+                    || !acc.getValue().equals( "BCL2_HUMAN" ) ) {
+                System.out.println( acc.toString() );
+                return false;
+            }
+            n.setName( "P10415" );
+            acc = SequenceDbWsTools.obtainSeqAccession( n );
+            if ( ( acc == null ) || !acc.getSource().equals( Source.UNIPROT.toString() )
+                    || !acc.getValue().equals( "P10415" ) ) {
+                System.out.println( acc.toString() );
+                return false;
+            }
+            n.setName( " P10415 " );
+            acc = SequenceDbWsTools.obtainSeqAccession( n );
+            if ( ( acc == null ) || !acc.getSource().equals( Source.UNIPROT.toString() )
+                    || !acc.getValue().equals( "P10415" ) ) {
+                System.out.println( acc.toString() );
+                return false;
+            }
+            n.setName( "_P10415|" );
+            acc = SequenceDbWsTools.obtainSeqAccession( n );
+            if ( ( acc == null ) || !acc.getSource().equals( Source.UNIPROT.toString() )
+                    || !acc.getValue().equals( "P10415" ) ) {
+                System.out.println( acc.toString() );
+                return false;
+            }
+            n.setName( "AY695820" );
+            acc = SequenceDbWsTools.obtainSeqAccession( n );
+            if ( ( acc == null ) || !acc.getSource().equals( Source.NCBI.toString() )
+                    || !acc.getValue().equals( "AY695820" ) ) {
+                System.out.println( acc.toString() );
+                return false;
+            }
+            n.setName( "_AY695820_" );
+            acc = SequenceDbWsTools.obtainSeqAccession( n );
+            if ( ( acc == null ) || !acc.getSource().equals( Source.NCBI.toString() )
+                    || !acc.getValue().equals( "AY695820" ) ) {
+                System.out.println( acc.toString() );
+                return false;
+            }
+            n.setName( "AAA59452" );
+            acc = SequenceDbWsTools.obtainSeqAccession( n );
+            if ( ( acc == null ) || !acc.getSource().equals( Source.NCBI.toString() )
+                    || !acc.getValue().equals( "AAA59452" ) ) {
+                System.out.println( acc.toString() );
+                return false;
+            }
+            n.setName( "_AAA59452_" );
+            acc = SequenceDbWsTools.obtainSeqAccession( n );
+            if ( ( acc == null ) || !acc.getSource().equals( Source.NCBI.toString() )
+                    || !acc.getValue().equals( "AAA59452" ) ) {
+                System.out.println( acc.toString() );
+                return false;
+            }
+            n.setName( "AAA59452.1" );
+            acc = SequenceDbWsTools.obtainSeqAccession( n );
+            if ( ( acc == null ) || !acc.getSource().equals( Source.NCBI.toString() )
+                    || !acc.getValue().equals( "AAA59452.1" ) ) {
+                System.out.println( acc.toString() );
+                return false;
+            }
+            n.setName( "_AAA59452.1_" );
+            acc = SequenceDbWsTools.obtainSeqAccession( n );
+            if ( ( acc == null ) || !acc.getSource().equals( Source.NCBI.toString() )
+                    || !acc.getValue().equals( "AAA59452.1" ) ) {
+                System.out.println( acc.toString() );
+                return false;
+            }
+            n.setName( "GI:94894583" );
+            acc = SequenceDbWsTools.obtainSeqAccession( n );
+            if ( ( acc == null ) || !acc.getSource().equals( Source.GI.toString() )
+                    || !acc.getValue().equals( "94894583" ) ) {
+                System.out.println( acc.toString() );
+                return false;
+            }
+        }
+        catch ( final Exception e ) {
+            return false;
+        }
+        return true;
+    }
+
+    private static boolean testSequenceDbWsTools2() {
+        try {
+            final PhylogenyNode n1 = new PhylogenyNode( "NP_001025424" );
+            SequenceDbWsTools.obtainSeqInformation( n1 );
+            if ( !n1.getNodeData().getSequence().getName().equals( "Bcl2" ) ) {
+                return false;
+            }
+            if ( !n1.getNodeData().getTaxonomy().getScientificName().equals( "Danio rerio" ) ) {
+                return false;
+            }
+            if ( !n1.getNodeData().getSequence().getAccession().getSource().equals( Source.REFSEQ.toString() ) ) {
+                return false;
+            }
+            if ( !n1.getNodeData().getSequence().getAccession().getValue().equals( "NP_001025424" ) ) {
+                return false;
+            }
+            final PhylogenyNode n2 = new PhylogenyNode( "NM_001030253" );
+            SequenceDbWsTools.obtainSeqInformation( n2 );
+            System.out.println( n2.toString() );
+            if ( !n2.getNodeData().getSequence().getName()
+                    .equals( "Danio rerio B-cell leukemia/lymphoma 2 (bcl2), mRNA" ) ) {
+                return false;
+            }
+            if ( !n2.getNodeData().getTaxonomy().getScientificName().equals( "Danio rerio" ) ) {
+                return false;
+            }
+            if ( !n2.getNodeData().getSequence().getAccession().getSource().equals( Source.REFSEQ.toString() ) ) {
+                return false;
+            }
+            if ( !n2.getNodeData().getSequence().getAccession().getValue().equals( "NM_001030253" ) ) {
+                return false;
+            }
+            final PhylogenyNode n3 = new PhylogenyNode( "NM_184234.2" );
+            SequenceDbWsTools.obtainSeqInformation( n3 );
+            System.out.println( "n=" + n3.toString() );
+            if ( !n3.getNodeData().getSequence().getName()
+                    .equals( "Homo sapiens RNA binding motif protein 39 (RBM39), transcript variant 1, mRNA" ) ) {
+                return false;
+            }
+            if ( !n3.getNodeData().getTaxonomy().getScientificName().equals( "Homo sapiens" ) ) {
+                return false;
+            }
+            if ( !n3.getNodeData().getSequence().getAccession().getSource().equals( Source.REFSEQ.toString() ) ) {
+                return false;
+            }
+            if ( !n3.getNodeData().getSequence().getAccession().getValue().equals( "NM_184234" ) ) {
+                return false;
+            }
+        }
+        catch ( final IOException e ) {
+            System.out.println();
+            System.out.println( "the following might be due to absence internet connection:" );
+            e.printStackTrace( System.out );
+            return true;
+        }
+        catch ( final Exception e ) {
+            e.printStackTrace();
+            return false;
+        }
+        return true;
+    }
+
     private static boolean testUniprotEntryRetrieval() {
         try {
             final SequenceDatabaseEntry entry = SequenceDbWsTools.obtainUniProtEntry( "P12345", 200 );
@@ -10818,6 +11050,12 @@ public final class Test {
             if ( !entry.getSequenceName().equals( "Aspartate aminotransferase, mitochondrial" ) ) {
                 return false;
             }
+            if ( !entry.getSequenceSymbol().equals( "mAspAT" ) ) {
+                return false;
+            }
+            if ( !entry.getGeneName().equals( "GOT2" ) ) {
+                return false;
+            }
             if ( !entry.getTaxonomyIdentifier().equals( "9986" ) ) {
                 return false;
             }