private static boolean testEbiEntryRetrieval() {
try {
- final SequenceDatabaseEntry entry = SequenceDbWsTools
- .obtainEmblEntry( new Accession( "AAK41263", Accession.Source.NCBI ) );
+ final SequenceDatabaseEntry entry = SequenceDbWsTools.obtainEntry( "AAK41263" );
if ( !entry.getAccession().equals( "AAK41263" ) ) {
System.out.println( entry.getAccession() );
return false;
System.out.println( entry.getTaxonomyIdentifier() );
return false;
}
- if ( !entry.getAnnotations().get( 0 ).getRefValue().equals( "3.2.1.33" ) ) {
- System.out.println( entry.getAnnotations().get( 0 ).getRefValue() );
+ if ( !entry.getAnnotations().first().getRefValue().equals( "3.2.1.33" ) ) {
+ System.out.println( entry.getAnnotations().first().getRefValue() );
return false;
}
- if ( !entry.getAnnotations().get( 0 ).getRefSource().equals( "EC" ) ) {
- System.out.println( entry.getAnnotations().get( 0 ).getRefSource() );
+ if ( !entry.getAnnotations().first().getRefSource().equals( "EC" ) ) {
+ System.out.println( entry.getAnnotations().first().getRefSource() );
+ return false;
+ }
+ if ( entry.getCrossReferences().size() != 5 ) {
+ return false;
+ }
+ //
+ final SequenceDatabaseEntry entry1 = SequenceDbWsTools.obtainEntry( "ABJ16409" );
+ if ( !entry1.getAccession().equals( "ABJ16409" ) ) {
+ return false;
+ }
+ if ( !entry1.getTaxonomyScientificName().equals( "Felis catus" ) ) {
+ System.out.println( entry1.getTaxonomyScientificName() );
+ return false;
+ }
+ if ( !entry1.getSequenceName().equals( "Felis catus (domestic cat) partial BCL2" ) ) {
+ System.out.println( entry1.getSequenceName() );
+ return false;
+ }
+ if ( !entry1.getTaxonomyIdentifier().equals( "9685" ) ) {
+ System.out.println( entry1.getTaxonomyIdentifier() );
+ return false;
+ }
+ if ( !entry1.getGeneName().equals( "BCL2" ) ) {
+ System.out.println( entry1.getGeneName() );
+ return false;
+ }
+ if ( entry1.getCrossReferences().size() != 6 ) {
+ return false;
+ }
+ //
+ final SequenceDatabaseEntry entry2 = SequenceDbWsTools.obtainEntry( "NM_184234" );
+ if ( !entry2.getAccession().equals( "NM_184234" ) ) {
+ return false;
+ }
+ if ( !entry2.getTaxonomyScientificName().equals( "Homo sapiens" ) ) {
+ System.out.println( entry2.getTaxonomyScientificName() );
+ return false;
+ }
+ if ( !entry2.getSequenceName()
+ .equals( "Homo sapiens RNA binding motif protein 39 (RBM39), transcript variant 1, mRNA" ) ) {
+ System.out.println( entry2.getSequenceName() );
+ return false;
+ }
+ if ( !entry2.getTaxonomyIdentifier().equals( "9606" ) ) {
+ System.out.println( entry2.getTaxonomyIdentifier() );
+ return false;
+ }
+ if ( !entry2.getGeneName().equals( "RBM39" ) ) {
+ System.out.println( entry2.getGeneName() );
+ return false;
+ }
+ if ( entry2.getCrossReferences().size() != 3 ) {
+ return false;
+ }
+ //
+ final SequenceDatabaseEntry entry3 = SequenceDbWsTools.obtainEntry( "HM043801" );
+ if ( !entry3.getAccession().equals( "HM043801" ) ) {
+ return false;
+ }
+ if ( !entry3.getTaxonomyScientificName().equals( "Bursaphelenchus xylophilus" ) ) {
+ System.out.println( entry3.getTaxonomyScientificName() );
+ return false;
+ }
+ if ( !entry3.getSequenceName().equals( "Bursaphelenchus xylophilus RAF gene, complete cds" ) ) {
+ System.out.println( entry3.getSequenceName() );
+ return false;
+ }
+ if ( !entry3.getTaxonomyIdentifier().equals( "6326" ) ) {
+ System.out.println( entry3.getTaxonomyIdentifier() );
+ return false;
+ }
+ if ( !entry3.getSequenceSymbol().equals( "RAF" ) ) {
+ System.out.println( entry3.getSequenceSymbol() );
+ return false;
+ }
+ if ( !ForesterUtil.isEmpty( entry3.getGeneName() ) ) {
+ return false;
+ }
+ if ( entry3.getCrossReferences().size() != 8 ) {
+ return false;
+ }
+ //
+ //
+ final SequenceDatabaseEntry entry4 = SequenceDbWsTools.obtainEntry( "AAA36557.1" );
+ if ( !entry4.getAccession().equals( "AAA36557" ) ) {
+ return false;
+ }
+ //
+ final SequenceDatabaseEntry entry5 = SequenceDbWsTools.obtainEntry( "M30539" );
+ if ( !entry5.getAccession().equals( "HM043801" ) ) {
return false;
}
}
package org.forester.ws.seqdb;
-import java.util.ArrayList;
import java.util.List;
+import java.util.SortedSet;
+import java.util.TreeSet;
import java.util.regex.Matcher;
import java.util.regex.Pattern;
final Pattern chromosome_PATTERN = Pattern.compile( "\\s+/chromosome=\"(\\w+)\"" );
final Pattern map_PATTERN = Pattern.compile( "\\s+/map=\"([\\w+\\.])\"" );
final Pattern gene_PATTERN = Pattern.compile( "\\s+/gene=\"(.+)\"" );
- final Pattern mim_xref_PATTERN = Pattern.compile( "\\s+/db_xref=\"MIM:(\\d+)\"" );
- final Pattern taxon_xref_PATTERN = Pattern.compile( "\\s+/db_xref=\"taxon:(\\d+)\"" );
- final Pattern interpro_PATTERN = Pattern.compile( "\\s+/db_xref=\"InterPro:(IP\\d+)\"" );
+ final Pattern mim_PATTERN = Pattern.compile( "\\s+/db_xref=\"MIM:(\\d+)\"" );
+ final Pattern taxon_PATTERN = Pattern.compile( "\\s+/db_xref=\"taxon:(\\d+)\"" );
+ final Pattern interpro_PATTERN = Pattern.compile( "\\s+/db_xref=\"InterPro:([A-Z0-9]+)\"" );
final Pattern uniprot_PATTERN = Pattern.compile( "\\s+/db_xref=\"UniProtKB/TrEMBL:(\\w+)\"" );
+ final Pattern hgnc_PATTERN = Pattern.compile( "\\s+/db_xref=\"HGNC:(\\d+)\"" );
+ final Pattern geneid_PATTERN = Pattern.compile( "\\s+/db_xref=\"GeneID:(\\d+)\"" );
final Pattern ec_PATTERN = Pattern.compile( "\\s+/EC_number=\"([\\.\\-\\d]+)\"" );
+ final Pattern product_PATTERN = Pattern.compile( "\\s+/product=\"(\\w{1,10})\"" );
final EbiDbEntry e = new EbiDbEntry();
final StringBuilder def = new StringBuilder();
boolean in_definition = false;
boolean in_source = false;
boolean in_gene = false;
boolean in_cds = false;
+ boolean in_mrna = false;
boolean in_protein = false;
for( final String line : lines ) {
if ( line.startsWith( "ACCESSION " ) ) {
in_source = false;
in_gene = false;
in_cds = false;
+ in_mrna = false;
in_protein = false;
// in_def = false;
}
in_source = true;
in_gene = false;
in_cds = false;
+ in_mrna = false;
in_protein = false;
}
if ( in_features && ( line.startsWith( " gene " ) || line.startsWith( "FT gene " ) ) ) {
in_source = false;
in_gene = true;
in_cds = false;
+ in_mrna = false;
in_protein = false;
}
if ( in_features && ( line.startsWith( " CDS " ) || line.startsWith( "FT CDS " ) ) ) {
in_source = false;
in_gene = false;
in_cds = true;
+ in_mrna = false;
in_protein = false;
}
if ( in_features && ( line.startsWith( " Protein " ) || line.startsWith( "FT Protein " ) ) ) {
in_source = false;
in_gene = false;
in_cds = false;
+ in_mrna = false;
in_protein = true;
}
+ if ( in_features && ( line.startsWith( " mRNA " ) || line.startsWith( "FT mRNA " ) ) ) {
+ in_source = false;
+ in_gene = false;
+ in_cds = false;
+ in_mrna = true;
+ in_protein = false;
+ }
if ( in_source ) {
- final Matcher m = taxon_xref_PATTERN.matcher( line );
+ final Matcher m = taxon_PATTERN.matcher( line );
if ( m.find() ) {
e.setTaxId( m.group( 1 ) );
}
}
- if ( in_protein || in_cds ) {
- final Matcher m = ec_PATTERN.matcher( line );
- if ( m.find() ) {
- e.addAnnotation( new Annotation( "EC", m.group( 1 ) ) );
+ if ( in_cds || in_gene ) {
+ final Matcher hgnc = hgnc_PATTERN.matcher( line );
+ if ( hgnc.find() ) {
+ e.addCrossReference( new Accession( hgnc.group( 1 ), "hgnc" ) );
+ }
+ final Matcher geneid = geneid_PATTERN.matcher( line );
+ if ( geneid.find() ) {
+ e.addCrossReference( new Accession( geneid.group( 1 ), "geneid" ) );
}
}
- if ( in_protein || in_cds || in_gene ) {
- final Matcher m = gene_PATTERN.matcher( line );
- if ( m.find() ) {
- e.setGeneName( m.group( 1 ) );
+ if ( in_protein || in_cds || in_gene || in_mrna ) {
+ final Matcher ec = ec_PATTERN.matcher( line );
+ if ( ec.find() ) {
+ e.addAnnotation( new Annotation( "EC", ec.group( 1 ) ) );
+ }
+ final Matcher gene = gene_PATTERN.matcher( line );
+ if ( gene.find() ) {
+ e.setGeneName( gene.group( 1 ) );
+ }
+ final Matcher uniprot = uniprot_PATTERN.matcher( line );
+ if ( uniprot.find() ) {
+ e.addCrossReference( new Accession( uniprot.group( 1 ), "uniprot" ) );
+ }
+ final Matcher interpro = interpro_PATTERN.matcher( line );
+ if ( interpro.find() ) {
+ e.addCrossReference( new Accession( interpro.group( 1 ), "interpro" ) );
+ }
+ final Matcher mim = mim_PATTERN.matcher( line );
+ if ( mim.find() ) {
+ e.addCrossReference( new Accession( mim.group( 1 ), "mim" ) );
+ }
+ final Matcher product = product_PATTERN.matcher( line );
+ if ( product.find() ) {
+ e.setSequenceSymbol( product.group( 1 ) );
}
}
}
}
// FIXME actually this is NCBI entry
//http://www.ebi.ac.uk/Tools/dbfetch/dbfetch/emb/AAR37336/
- private String _pa;
- private String _de;
- private String _os;
- private String _tax_id;
- private String _symbol;
- private String _provider;
- private List<Accession> _cross_references;
- private List<Annotation> _annotations;
- private String _gene_name;
+ private String _pa;
+ private String _de;
+ private String _os;
+ private String _tax_id;
+ private String _symbol;
+ private String _provider;
+ private SortedSet<Accession> _cross_references;
+ private SortedSet<Annotation> _annotations;
+ private String _gene_name;
// TODO PUBMED 15798186
//TODO (FEATURES)
private void addCrossReference( final Accession accession ) {
if ( _cross_references == null ) {
- _cross_references = new ArrayList<Accession>();
+ _cross_references = new TreeSet<Accession>();
}
System.out.println( "XREF ADDED: " + accession );
_cross_references.add( accession );
}
@Override
- public List<Accession> getCrossReferences() {
+ public SortedSet<Accession> getCrossReferences() {
return _cross_references;
}
}
@Override
- public List<GoTerm> getGoTerms() {
+ public SortedSet<GoTerm> getGoTerms() {
return null;
}
return _symbol;
}
+ private void setSequenceSymbol( String symbol ) {
+ _symbol = symbol;
+ }
+
@Override
public String getTaxonomyIdentifier() {
return _tax_id;
}
@Override
- public List<Annotation> getAnnotations() {
+ public SortedSet<Annotation> getAnnotations() {
return _annotations;
}
private void addAnnotation( final Annotation annotation ) {
if ( _annotations == null ) {
- _annotations = new ArrayList<Annotation>();
+ _annotations = new TreeSet<Annotation>();
}
_annotations.add( annotation );
}
package org.forester.ws.seqdb;
-import java.util.List;
+import java.util.SortedSet;
import org.forester.go.GoTerm;
import org.forester.phylogeny.data.Accession;
public String getGeneName();
- public List<GoTerm> getGoTerms();
+ public SortedSet<GoTerm> getGoTerms();
- public List<Annotation> getAnnotations();
+ public SortedSet<Annotation> getAnnotations();
public String getProvider();
public boolean isEmpty();
- public List<Accession> getCrossReferences();
+ public SortedSet<Accession> getCrossReferences();
}
\ No newline at end of file
public final class SequenceDbWsTools {
- public final static String EMBL_REFSEQ = "http://www.ebi.ac.uk/Tools/dbfetch/dbfetch?db=REFSEQ&style=raw&id=";
- public final static String EMBL_GENBANK = "http://www.ebi.ac.uk/Tools/dbfetch/dbfetch?db=GENBANK&style=raw&id=";
public final static String BASE_UNIPROT_URL = "http://www.uniprot.org/";
+ public final static int DEFAULT_LINES_TO_RETURN = 4000;
//public final static String EMBL_DBS_EMBL = "embl";
public final static String EMBL_DBS_REFSEQ_N = "refseqn";
public final static String EMBL_DBS_REFSEQ_P = "refseqp";
+ public final static String EMBL_GENBANK = "http://www.ebi.ac.uk/Tools/dbfetch/dbfetch?db=GENBANK&style=raw&id=";
+ public final static String EMBL_REFSEQ = "http://www.ebi.ac.uk/Tools/dbfetch/dbfetch?db=REFSEQ&style=raw&id=";
private final static boolean DEBUG = true;
private final static String URL_ENC = "UTF-8";
- public final static int DEFAULT_LINES_TO_RETURN = 4000;
-
- final static String extractFrom( final String target, final String a ) {
- final int i_a = target.indexOf( a );
- return target.substring( i_a + a.length() ).trim();
- }
-
- final static String extractFromTo( final String target, final String a, final String b ) {
- final int i_a = target.indexOf( a );
- final int i_b = target.indexOf( b );
- if ( ( i_a < 0 ) || ( i_b < i_a ) ) {
- throw new IllegalArgumentException( "attempt to extract from \"" + target + "\" between \"" + a
- + "\" and \"" + b + "\"" );
- }
- return target.substring( i_a + a.length(), i_b ).trim();
- }
-
- final static String extractTo( final String target, final String b ) {
- final int i_b = target.indexOf( b );
- return target.substring( 0, i_b ).trim();
- }
public static List<UniProtTaxonomy> getTaxonomiesFromCommonNameStrict( final String cn,
final int max_taxonomies_return )
return null;
}
+ public static SequenceDatabaseEntry obtainEmblEntry( final Accession acc ) throws IOException {
+ return obtainEmblEntry( acc, DEFAULT_LINES_TO_RETURN );
+ }
+
public static SequenceDatabaseEntry obtainEmblEntry( final Accession acc, final int max_lines_to_return )
throws IOException {
final List<String> lines = queryEmblDb( acc, max_lines_to_return );
return EbiDbEntry.createInstanceFromPlainTextForRefSeq( lines );
}
- public static SequenceDatabaseEntry obtainEmblEntry( final Accession acc ) throws IOException {
- return obtainEmblEntry( acc, DEFAULT_LINES_TO_RETURN );
+ public static SequenceDatabaseEntry obtainEntry( final String acc_str ) throws IOException {
+ if ( ForesterUtil.isEmpty( acc_str ) ) {
+ throw new IllegalArgumentException( "cannot not extract sequence db accessor from null or empty string" );
+ }
+ final Accession acc = SequenceAccessionTools.parseAccessorFromString( acc_str );
+ if ( acc == null ) {
+ throw new IllegalArgumentException( "could not extract acceptable sequence db accessor from \"" + acc_str
+ + "\"" );
+ }
+ if ( acc.getSource().equals( Source.REFSEQ.toString() ) || acc.getSource().equals( Source.EMBL.toString() )
+ || acc.getSource().equals( Source.NCBI.toString() ) ) {
+ return obtainEmblEntry( acc, DEFAULT_LINES_TO_RETURN );
+ }
+ else if ( acc.getSource().equals( Source.UNIPROT.toString() ) ) {
+ return obtainUniProtEntry( acc.getValue(), DEFAULT_LINES_TO_RETURN );
+ }
+ else {
+ throw new IllegalArgumentException( "don't know how to handle request for source \"" + acc.getSource()
+ + "\"" );
+ }
}
- public final static Accession obtainSeqAccession( final PhylogenyNode node ) {
- Accession acc = SequenceAccessionTools.obtainFromSeqAccession( node );
- if ( !isAccessionAcceptable( acc ) ) {
- acc = SequenceAccessionTools.obtainAccessorFromDataFields( node );
- }
- return acc;
+ public static SequenceDatabaseEntry obtainRefSeqEntryFromEmbl( final Accession acc ) throws IOException {
+ return obtainRefSeqEntryFromEmbl( acc, DEFAULT_LINES_TO_RETURN );
}
public static SequenceDatabaseEntry obtainRefSeqEntryFromEmbl( final Accession acc, final int max_lines_to_return )
return EbiDbEntry.createInstanceFromPlainTextForRefSeq( lines );
}
- public static SequenceDatabaseEntry obtainRefSeqEntryFromEmbl( final Accession acc ) throws IOException {
- return obtainRefSeqEntryFromEmbl( acc, DEFAULT_LINES_TO_RETURN );
+ public final static Accession obtainSeqAccession( final PhylogenyNode node ) {
+ Accession acc = SequenceAccessionTools.obtainFromSeqAccession( node );
+ if ( !isAccessionAcceptable( acc ) ) {
+ acc = SequenceAccessionTools.obtainAccessorFromDataFields( node );
+ }
+ return acc;
}
public final static void obtainSeqInformation( final boolean allow_to_set_taxonomic_data,
obtainSeqInformation( allow_to_set_taxonomic_data, DEFAULT_LINES_TO_RETURN, not_found, node );
}
- public final static void obtainSeqInformation( final PhylogenyNode node ) throws IOException {
- obtainSeqInformation( true, DEFAULT_LINES_TO_RETURN, new TreeSet<String>(), node );
- }
-
public final static SortedSet<String> obtainSeqInformation( final Phylogeny phy,
final boolean ext_nodes_only,
final boolean allow_to_set_taxonomic_data,
return not_found;
}
- public static SequenceDatabaseEntry obtainUniProtEntry( final String query, final int max_lines_to_return )
- throws IOException {
- final List<String> lines = queryUniprot( "uniprot/" + query + ".txt", max_lines_to_return );
- return UniProtEntry.createInstanceFromPlainText( lines );
+ public final static void obtainSeqInformation( final PhylogenyNode node ) throws IOException {
+ obtainSeqInformation( true, DEFAULT_LINES_TO_RETURN, new TreeSet<String>(), node );
}
public static SequenceDatabaseEntry obtainUniProtEntry( final String query ) throws IOException {
return obtainUniProtEntry( query, DEFAULT_LINES_TO_RETURN );
}
+ public static SequenceDatabaseEntry obtainUniProtEntry( final String query, final int max_lines_to_return )
+ throws IOException {
+ final List<String> lines = queryUniprot( "uniprot/" + query + ".txt", max_lines_to_return );
+ return UniProtEntry.createInstanceFromPlainText( lines );
+ }
+
public static List<String> queryDb( final String query, int max_lines_to_return, final String base_url )
throws IOException {
if ( ForesterUtil.isEmpty( query ) ) {
return result;
}
- public static List<String> queryEmblDbForRefSeqEntry( final Accession id, final int max_lines_to_return )
- throws IOException {
- final StringBuilder url_sb = new StringBuilder();
- url_sb.append( EMBL_REFSEQ );
- return queryDb( id.getValue(), max_lines_to_return, url_sb.toString() );
- }
-
public static List<String> queryEmblDb( final Accession id, final int max_lines_to_return ) throws IOException {
final StringBuilder url_sb = new StringBuilder();
// url_sb.append( BASE_EMBL_DB_URL );
return queryDb( id.getValue(), max_lines_to_return, url_sb.toString() );
}
+ public static List<String> queryEmblDbForRefSeqEntry( final Accession id, final int max_lines_to_return )
+ throws IOException {
+ final StringBuilder url_sb = new StringBuilder();
+ url_sb.append( EMBL_REFSEQ );
+ return queryDb( id.getValue(), max_lines_to_return, url_sb.toString() );
+ }
+
public static List<String> queryUniprot( final String query, final int max_lines_to_return ) throws IOException {
return queryDb( query, max_lines_to_return, BASE_UNIPROT_URL );
}
+ final static String extractFrom( final String target, final String a ) {
+ final int i_a = target.indexOf( a );
+ return target.substring( i_a + a.length() ).trim();
+ }
+
+ final static String extractFromTo( final String target, final String a, final String b ) {
+ final int i_a = target.indexOf( a );
+ final int i_b = target.indexOf( b );
+ if ( ( i_a < 0 ) || ( i_b < i_a ) ) {
+ throw new IllegalArgumentException( "attempt to extract from \"" + target + "\" between \"" + a
+ + "\" and \"" + b + "\"" );
+ }
+ return target.substring( i_a + a.length(), i_b ).trim();
+ }
+
+ final static String extractTo( final String target, final String b ) {
+ final int i_b = target.indexOf( b );
+ return target.substring( 0, i_b ).trim();
+ }
+
private static void addDataFromDbToNode( final boolean allow_to_set_taxonomic_data,
final int lines_to_return,
final SortedSet<String> not_found,
package org.forester.ws.seqdb;
-import java.util.ArrayList;
import java.util.List;
+import java.util.SortedSet;
+import java.util.TreeSet;
import java.util.regex.Matcher;
import java.util.regex.Pattern;
public final static Pattern PharmGKB_PATTERN = Pattern.compile( "PharmGKB;\\s+([0-9A-Z]+);" );
public final static Pattern Reactome_PATTERN = Pattern.compile( "Reactome;\\s+([0-9A-Z]+);\\s+([^\\.]+)" );
private String _ac;
- private ArrayList<Accession> _cross_references;
+ private SortedSet<Accession> _cross_references;
private String _gene_name;
- private List<GoTerm> _go_terms;
+ private SortedSet<GoTerm> _go_terms;
private String _name;
private String _os_scientific_name;
private String _symbol;
}
@Override
- public List<Accession> getCrossReferences() {
+ public SortedSet<Accession> getCrossReferences() {
return _cross_references;
}
}
@Override
- public List<GoTerm> getGoTerms() {
+ public SortedSet<GoTerm> getGoTerms() {
return _go_terms;
}
private void addCrossReference( final Accession accession ) {
if ( _cross_references == null ) {
- _cross_references = new ArrayList<Accession>();
+ _cross_references = new TreeSet<Accession>();
}
_cross_references.add( accession );
}
private void addGoTerm( final BasicGoTerm g ) {
if ( _go_terms == null ) {
- _go_terms = new ArrayList<GoTerm>();
+ _go_terms = new TreeSet<GoTerm>();
}
_go_terms.add( g );
}
}
@Override
- public List<Annotation> getAnnotations() {
+ public SortedSet<Annotation> getAnnotations() {
return null;
}
}
git://source.jalview.org / jalview.git / commitdiff