package org.forester.ws.seqdb;
import java.util.List;
+import java.util.SortedSet;
+import java.util.TreeSet;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
import org.forester.go.GoTerm;
import org.forester.phylogeny.data.Accession;
+import org.forester.phylogeny.data.Annotation;
+import org.forester.sequence.MolecularSequence;
import org.forester.util.ForesterUtil;
public final class EbiDbEntry implements SequenceDatabaseEntry {
+ private SortedSet<Annotation> _annotations;
+ private String _chromosome;
+ private SortedSet<Accession> _cross_references;
+ private String _de;
+ private String _gene_name;
+ private String _map;
+ private String _os;
// 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 String _pa;
+ private String _provider;
+ private String _symbol;
+ private String _tax_id;
// TODO PUBMED 15798186
//TODO (FEATURES)
// /db_xref="MIM:604739"
// /db_xref="InterPro:IPR002475"
// /product="Bcl-2"
- // /protein_id="NP_909122.1"
// /db_xref="UniProtKB/TrEMBL:Q5J7V1" <- reparse?
//
// Protein
recorded"
/note="N-acetylalanine; propagated from
UniProtKB/Swiss-Prot (Q14498.2); acetylation site"
- misc_feature 692..694
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /experiment="experimental evidence, no additional details
- recorded"
- /note="Phosphotyrosine; propagated from
- UniProtKB/Swiss-Prot (Q14498.2); phosphorylation site"
- misc_feature 698..700
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /experiment="experimental evidence, no additional details
- recorded"
- /note="Phosphoserine; propagated from UniProtKB/Swiss-Prot
- (Q14498.2); phosphorylation site"
- misc_feature 707..709
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /experiment="experimental evidence, no additional details
- recorded"
- /note="Phosphoserine; propagated from UniProtKB/Swiss-Prot
- (Q14498.2); phosphorylation site"
- misc_feature 815..817
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /experiment="experimental evidence, no additional details
- recorded"
- /note="Phosphoserine; propagated from UniProtKB/Swiss-Prot
- (Q14498.2); phosphorylation site"
- misc_feature 845..847
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /experiment="experimental evidence, no additional details
- recorded"
- /note="Phosphothreonine; propagated from
- UniProtKB/Swiss-Prot (Q14498.2); phosphorylation site"
- misc_feature 1280..1627
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /inference="non-experimental evidence, no additional
- details recorded"
- /note="propagated from UniProtKB/Swiss-Prot (Q14498.2);
- Region: Interaction with JUN (By similarity)"
- misc_feature 1280..1474
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /inference="non-experimental evidence, no additional
- details recorded"
- /note="propagated from UniProtKB/Swiss-Prot (Q14498.2);
- Region: Activating domain (By similarity)"
- misc_feature 1409..1411
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /experiment="experimental evidence, no additional details
- recorded"
- /note="Phosphoserine; propagated from UniProtKB/Swiss-Prot
- (Q14498.2); phosphorylation site"
- misc_feature 1418..1420
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /experiment="experimental evidence, no additional details
- recorded"
- /note="Phosphoserine; propagated from UniProtKB/Swiss-Prot
- (Q14498.2); phosphorylation site"
- misc_feature 1430..1432
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /experiment="experimental evidence, no additional details
- recorded"
- /note="Phosphoserine; propagated from UniProtKB/Swiss-Prot
- (Q14498.2); phosphorylation site"
- misc_feature 1472..1627
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /inference="non-experimental evidence, no additional
- details recorded"
- /note="propagated from UniProtKB/Swiss-Prot (Q14498.2);
- Region: Interaction with ESR1 and ESR2 (By similarity)"
- misc_feature 1625..1999
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /inference="non-experimental evidence, no additional
- details recorded"
- /note="propagated from UniProtKB/Swiss-Prot (Q14498.2);
- Region: Interaction with NCOA6 (By similarity)"
+
exon 461..510
/gene="RBM39"
/gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
/inference="alignment:Splign:1.39.8"
- exon 511..705
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /inference="alignment:Splign:1.39.8"
- exon 706..771
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /inference="alignment:Splign:1.39.8"
- exon 772..825
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /inference="alignment:Splign:1.39.8"
- exon 826..943
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /inference="alignment:Splign:1.39.8"
- exon 944..1096
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /inference="alignment:Splign:1.39.8"
- exon 1097..1234
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /inference="alignment:Splign:1.39.8"
- exon 1235..1300
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /inference="alignment:Splign:1.39.8"
- exon 1301..1505
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /inference="alignment:Splign:1.39.8"
- exon 1506..1583
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /inference="alignment:Splign:1.39.8"
- exon 1584..1634
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /inference="alignment:Splign:1.39.8"
- exon 1635..1716
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /inference="alignment:Splign:1.39.8"
- exon 1717..1822
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /inference="alignment:Splign:1.39.8"
- exon 1823..1901
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /inference="alignment:Splign:1.39.8"
+
exon 1902..2874
/gene="RBM39"
/gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
/gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
/standard_name="REN58785"
/db_xref="UniSTS:383585"
- STS 2349..2590
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /standard_name="REN58784"
- /db_xref="UniSTS:383584"
- STS 2450..2669
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /standard_name="RH69003"
- /db_xref="UniSTS:85360"
- STS 2579..2828
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /standard_name="REN58783"
- /db_xref="UniSTS:383583"
- STS 2639..2728
- /gene="RBM39"
- /gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
- /standard_name="RH67917"
- /db_xref="UniSTS:84037"
+
polyA_signal 2851..2856
/gene="RBM39"
/gene_synonym="CAPER; CAPERalpha; FSAP59; HCC1; RNPC2"
121 ttcagggagc ttgtcgacgc tgtcgcaggg gtggatcctg agctgccgaa gccgccgtcc
181 tgctctcccg cgtgggcttc tctaattcca ttgttttttt tagattctct cgggcctagc
241 cgtccttgga acccgatatt cgggctgggc ggttccgcgg cctgggccta ggggcttaac
- 301 agtagcaaca gaagcggcgg cggcggcagc agcagcagca gcagcagcaa tctcttcccg
- 361 aacacgagca ccacaggcgc ccgaaggccg gaacaggcgt ttagagaaaa tggcagacga
- 421 tattgatatt gaagcaatgc ttgaggctcc ttacaagaag gatgagaaca agttgagcag
- 481 tgccaacggc catgaagaac gtagcaaaaa gaggaaaaaa agcaagagca gaagtcgtag
- 541 tcatgaacga aagagaagca aaagtaagga acggaagcga agtagagaca gagaaaggaa
- 601 aaagagcaaa agccgtgaaa gaaagcgaag tagaagcaaa gagaggcgac ggagccgctc
- 661 aagaagtcga gatcgaagat ttagaggccg ctacagaagt ccttactccg gaccaaaatt
- 721 taacagtgcc atccgaggaa agattgggtt gcctcatagc atcaaattaa gcagacgacg
- 781 ttcccgaagc aaaagtccat tcagaaaaga caagagccct gtgagagaac ctattgataa
- 841 tttaactcct gaggaaagag atgcaaggac agtcttctgt atgcagctgg cggcaagaat
- 901 tcgaccaagg gatttggaag agtttttctc tacagtagga aaggttcgag atgtgaggat
- 961 gatttctgac agaaattcaa gacgttccaa aggaattgct tatgtggagt tcgtcgatgt
- 1021 tagctcagtg cctctagcaa taggattaac tggccaacga gttttaggcg tgccaatcat
- 1081 agtacaggca tcacaggcag aaaaaaacag agctgcagca atggcaaaca atttacaaaa
- 1141 gggaagtgct ggacctatga ggctttatgt gggctcatta cacttcaaca taactgaaga
- 1201 tatgcttcgt gggatctttg agccttttgg aagaattgaa agtatccagc tgatgatgga
- 1261 cagtgaaact ggtcgatcca agggatatgg atttattaca ttttctgact cagaatgtgc
- 1321 caaaaaggct ttggaacaac ttaatggatt tgaactagca ggaagaccaa tgaaagttgg
- 1381 tcatgttact gaacgtactg atgcttcgag tgctagttca tttttggaca gtgatgaact
- 1441 ggaaaggact ggaattgatt tgggaacaac tggtcgtctt cagttaatgg caagacttgc
- 1501 agagggtaca ggtttgcaga ttccgccagc agcacagcaa gctctacaga tgagtggctc
- 1561 tttggcattt ggtgctgtgg cagaattctc ttttgttata gatttgcaaa caagactttc
- 1621 ccagcagact gaagcttcag ctttagctgc agctgcctct gttcagccac ttgcaacaca
- 1681 atgtttccaa ctctctaaca tgtttaaccc tcaaacagaa gaagaagttg gatgggatac
- 1741 cgagattaag gatgatgtga ttgaagaatg taataaacat ggaggagtta ttcatattta
- 1801 tgttgacaaa aattcagctc agggcaatgt gtatgtgaag tgcccatcaa ttgctgcagc
- 1861 tattgctgct gtcaatgcat tgcatggcag gtggtttgct ggtaaaatga taacagcagc
- 1921 atatgtacct cttccaactt accacaacct gtttcctgat tctatgacag caacacagct
- 1981 actggttcca agtagacgat gaaggaagat atagtccctt atgtatatag ctttttttct
- 2041 ttcttgagaa ttcatcttga gttatctttt atttagataa aaataaagag gcaaggatct
- 2101 actgtcattt gtatgcaatt tcctgttacc ttgaaaaaat aaaaatgtta acaggaatgc
- 2161 agtgtgctca ttctccctaa atagtaaatc ccactgtata caaaactgtt ctcttgttct
- 2221 gccttttaaa atgttcatgt agaaaattaa tgaactatag gaatagctct aggagaacaa
- 2281 atgtgctttc tgtaaaaagg cagaccaggg atgtaatgtt tttaatgttt cagaagccta
- 2341 actttttaca cagtggttac atttcacatt tcactaatgt tgatatttgg ctgatggttg
- 2401 agcagtttct gaaatacaca tttagtgtat ggaaatacaa gacagctaaa gggctgtttg
- 2461 gttagcatct catcttgcat tctgatcaat tggcaagaaa gggagatttc aaaattatat
- 2521 ttcttgatgg tatcttttca attaatgtat ctgtaaaagt ttctttgtaa atactatgtg
- 2581 ttctggtgtg tcttaaaatt ccaaacaaaa tgatccctgc atttcctgaa gatgtttaaa
- 2641 cgtgagagtc tggtaggcaa agcagtctga gaaagaaata ggaaatgcag aaataggttt
- 2701 tgtctggttg catataatct ttgctctttt taagctctgt gagctctgaa atatattttt
- 2761 gggttacttc agtgtgtttg acaagacagc ttgatatttc tatcaaacaa atgactttca
- 2821 tattgcaaca atctttgtaa gaaccactca aataaaagtc tcttaaaaag gccaaaaaaa
- 2881 a
+
*/
throw new CloneNotSupportedException();
}
+ @Override
+ public String getAccession() {
+ return _pa;
+ }
+
+ @Override
+ public SortedSet<Annotation> getAnnotations() {
+ return _annotations;
+ }
+
+ @Override
+ public String getChromosome() {
+ return _chromosome;
+ }
+
+ @Override
+ public SortedSet<Accession> getCrossReferences() {
+ return _cross_references;
+ }
+
+ @Override
+ public String getGeneName() {
+ return _gene_name;
+ }
+
+ @Override
+ public SortedSet<GoTerm> getGoTerms() {
+ return null;
+ }
+
+ @Override
+ public String getMap() {
+ return _map;
+ }
+
+ @Override
+ public String getProvider() {
+ return _provider;
+ }
+
+ @Override
+ public String getSequenceName() {
+ return _de;
+ }
+
+ @Override
+ public String getSequenceSymbol() {
+ return _symbol;
+ }
+
+ @Override
+ public String getTaxonomyIdentifier() {
+ return _tax_id;
+ }
+
+ @Override
+ public String getTaxonomyScientificName() {
+ return _os;
+ }
+
+ @Override
+ public boolean isEmpty() {
+ return ( ForesterUtil.isEmpty( getAccession() ) && ForesterUtil.isEmpty( getSequenceName() )
+ && ForesterUtil.isEmpty( getTaxonomyScientificName() )
+ && ForesterUtil.isEmpty( getTaxonomyIdentifier() ) && ForesterUtil.isEmpty( getSequenceSymbol() ) );
+ }
+
+ public void setProvider( final String provider ) {
+ _provider = provider;
+ }
+
+ private void addAnnotation( final Annotation annotation ) {
+ if ( _annotations == null ) {
+ _annotations = new TreeSet<Annotation>();
+ }
+ _annotations.add( annotation );
+ }
+
+ private void addCrossReference( final Accession accession ) {
+ if ( _cross_references == null ) {
+ _cross_references = new TreeSet<Accession>();
+ }
+ System.out.println( "XREF ADDED: " + accession );
+ _cross_references.add( accession );
+ }
+
+ private void setAccession( final String pa ) {
+ if ( _pa == null ) {
+ _pa = pa;
+ }
+ }
+
+ private void setChromosome( final String chromosome ) {
+ _chromosome = chromosome;
+ }
+
+ private void setGeneName( final String gene_name ) {
+ if ( _gene_name == null ) {
+ _gene_name = gene_name;
+ }
+ }
+
+ private void setMap( final String map ) {
+ _map = map;
+ }
+
+ private void setSequenceName( final String rec_name ) {
+ if ( _de == null ) {
+ _de = rec_name;
+ }
+ }
+
+ private void setSequenceSymbol( final String symbol ) {
+ _symbol = symbol;
+ }
+
+ private void setTaxId( final String tax_id ) {
+ if ( _tax_id == null ) {
+ _tax_id = tax_id;
+ }
+ }
+
+ private void setTaxonomyScientificName( final String os ) {
+ if ( _os == null ) {
+ _os = os;
+ }
+ }
+
+ // public static SequenceDatabaseEntry createInstanceFromPlainText( final List<String> lines ) {
+ // final EbiDbEntry e = new EbiDbEntry();
+ // for( final String line : lines ) {
+ // if ( line.startsWith( "PA" ) ) {
+ // e.setPA( SequenceDbWsTools.extractFrom( line, "PA" ) );
+ // }
+ // else if ( line.startsWith( "DE" ) ) {
+ // e.setDe( SequenceDbWsTools.extractFrom( line, "DE" ) );
+ // }
+ // else if ( line.startsWith( "OS" ) ) {
+ // if ( line.indexOf( "(" ) > 0 ) {
+ // e.setOs( SequenceDbWsTools.extractFromTo( line, "OS", "(" ) );
+ // }
+ // else {
+ // e.setOs( SequenceDbWsTools.extractFrom( line, "OS" ) );
+ // }
+ // }
+ // else if ( line.startsWith( "OX" ) ) {
+ // if ( line.indexOf( "NCBI_TaxID=" ) > 0 ) {
+ // e.setTaxId( SequenceDbWsTools.extractFromTo( line, "NCBI_TaxID=", ";" ) );
+ // }
+ // }
+ // }
+ // return e;
+ // }
public static SequenceDatabaseEntry createInstanceFromPlainTextForRefSeq( final List<String> lines ) {
+ final Pattern X_PATTERN = Pattern.compile( "^[A-Z]+" );
+ 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_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/[A-Za-z-]*:(\\w+)\"" );
+ final Pattern hgnc_PATTERN = Pattern.compile( "\\s+/db_xref=\"[A-Z:]*HGNC:(\\d+)\"" );
+ final Pattern geneid_PATTERN = Pattern.compile( "\\s+/db_xref=\"GeneID:(\\d+)\"" );
+ final Pattern pdb_PATTERN = Pattern.compile( "\\s+/db_xref=\"PDB:([A-Z0-9]+)\"" );
+ 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_def = false;
+ boolean in_definition = false;
+ boolean in_features = 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 ) {
- // System.out.println( "-" + line );
- if ( line.startsWith( "ACCESSION" ) ) {
- e.setPA( SequenceDbWsTools.extractFrom( line, "ACCESSION" ) );
- in_def = false;
+ if ( line.startsWith( "ACCESSION " ) ) {
+ e.setAccession( SequenceDbWsTools.extractFrom( line, "ACCESSION" ) );
+ in_definition = false;
}
- else if ( line.startsWith( "DEFINITION" ) ) {
+ else if ( line.startsWith( "ID " ) ) {
+ e.setAccession( SequenceDbWsTools.extractFromTo( line, "ID", ";" ) );
+ in_definition = false;
+ }
+ else if ( line.startsWith( "DEFINITION " ) || ( line.startsWith( "DE " ) ) ) {
+ boolean definiton = false;
+ if ( line.startsWith( "DEFINITION " ) ) {
+ definiton = true;
+ }
if ( line.indexOf( "[" ) > 0 ) {
- def.append( SequenceDbWsTools.extractFromTo( line, "DEFINITION", "[" ) );
+ if ( definiton ) {
+ x( def, ( SequenceDbWsTools.extractFromTo( line, "DEFINITION", "[" ) ) );
+ }
+ else {
+ x( def, ( SequenceDbWsTools.extractFromTo( line, "DE", "[" ) ) );
+ }
}
else if ( line.indexOf( "." ) > 0 ) {
- def.append( SequenceDbWsTools.extractFromTo( line, "DEFINITION", "." ) );
+ if ( definiton ) {
+ x( def, ( SequenceDbWsTools.extractFromTo( line, "DEFINITION", "." ) ) );
+ }
+ else {
+ x( def, ( SequenceDbWsTools.extractFromTo( line, "DE", "." ) ) );
+ }
+ }
+ else {
+ if ( definiton ) {
+ x( def, ( SequenceDbWsTools.extractFrom( line, "DEFINITION" ) ) );
+ }
+ else {
+ x( def, ( SequenceDbWsTools.extractFrom( line, "DE" ) ) );
+ }
+ }
+ if ( definiton ) {
+ in_definition = true;
+ }
+ }
+ else if ( line.startsWith( " ORGANISM " ) ) {
+ if ( line.indexOf( "(" ) > 0 ) {
+ e.setTaxonomyScientificName( SequenceDbWsTools.extractFromTo( line, " ORGANISM", "(" ) );
}
else {
- def.append( SequenceDbWsTools.extractFrom( line, "DEFINITION" ) );
+ e.setTaxonomyScientificName( SequenceDbWsTools.extractFrom( line, " ORGANISM" ) );
}
- in_def = true;
+ // in_def = false;
}
- else if ( line.startsWith( "SOURCE" ) ) {
+ else if ( line.startsWith( "OS " ) ) {
if ( line.indexOf( "(" ) > 0 ) {
- e.setOs( SequenceDbWsTools.extractFromTo( line, "SOURCE", "(" ) );
+ e.setTaxonomyScientificName( SequenceDbWsTools.extractFromTo( line, "OS", "(" ) );
}
else {
- e.setOs( SequenceDbWsTools.extractFrom( line, "SOURCE" ) );
+ e.setTaxonomyScientificName( SequenceDbWsTools.extractFrom( line, "OS" ) );
}
- in_def = false;
}
- else if ( line.startsWith( " " ) && in_def ) {
+ else if ( line.startsWith( " " ) && in_definition ) {
def.append( " " );
if ( line.indexOf( "[" ) > 0 ) {
def.append( SequenceDbWsTools.extractTo( line, "[" ) );
}
}
else {
- in_def = false;
+ in_definition = false;
}
- }
- if ( def.length() > 0 ) {
- e.setDe( def.toString().trim() );
- }
- return e;
- }
-
- public static SequenceDatabaseEntry createInstanceFromPlainText( final List<String> lines ) {
- final EbiDbEntry e = new EbiDbEntry();
- for( final String line : lines ) {
- if ( line.startsWith( "PA" ) ) {
- e.setPA( SequenceDbWsTools.extractFrom( line, "PA" ) );
+ if ( !line.startsWith( "FT " ) && X_PATTERN.matcher( line ).find() ) {
+ in_features = false;
+ in_source = false;
+ in_gene = false;
+ in_cds = false;
+ in_mrna = false;
+ in_protein = false;
+ // in_def = false;
}
- else if ( line.startsWith( "DE" ) ) {
- e.setDe( SequenceDbWsTools.extractFrom( line, "DE" ) );
+ if ( line.startsWith( "FEATURES " ) || line.startsWith( "FT " ) ) {
+ in_features = true;
}
- else if ( line.startsWith( "OS" ) ) {
- if ( line.indexOf( "(" ) > 0 ) {
- e.setOs( SequenceDbWsTools.extractFromTo( line, "OS", "(" ) );
+ if ( in_features && ( line.startsWith( " source " ) || line.startsWith( "FT source " ) ) ) {
+ 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 ti = taxon_PATTERN.matcher( line );
+ if ( ti.find() ) {
+ e.setTaxId( ti.group( 1 ) );
}
- else {
- e.setOs( SequenceDbWsTools.extractFrom( line, "OS" ) );
+ final Matcher chr = chromosome_PATTERN.matcher( line );
+ if ( chr.find() ) {
+ e.setChromosome( chr.group( 1 ) );
+ }
+ final Matcher map = map_PATTERN.matcher( line );
+ if ( map.find() ) {
+ e.setMap( map.group( 1 ) );
}
}
- else if ( line.startsWith( "OX" ) ) {
- if ( line.indexOf( "NCBI_TaxID=" ) > 0 ) {
- e.setTaxId( SequenceDbWsTools.extractFromTo( line, "NCBI_TaxID=", ";" ) );
+ 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 || 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 ) );
+ }
+ final Matcher pdb = pdb_PATTERN.matcher( line );
+ if ( pdb.find() ) {
+ e.addCrossReference( new Accession( pdb.group( 1 ), "pdb" ) );
}
}
}
- return e;
- }
-
- @Override
- public String getAccession() {
- return _pa;
- }
-
- private void setPA( final String pa ) {
- if ( _pa == null ) {
- _pa = pa;
- }
- }
-
- @Override
- public String getSequenceName() {
- return _de;
- }
-
- private void setDe( final String rec_name ) {
- if ( _de == null ) {
- _de = rec_name;
- }
- }
-
- @Override
- public String getTaxonomyScientificName() {
- return _os;
- }
-
- private void setOs( final String os ) {
- if ( _os == null ) {
- _os = os;
+ if ( def.length() > 0 ) {
+ e.setSequenceName( def.toString().trim() );
}
+ return e;
}
- @Override
- public String getTaxonomyIdentifier() {
- return _tax_id;
- }
-
- private void setTaxId( final String tax_id ) {
- if ( _tax_id == null ) {
- _tax_id = tax_id;
+ private static void x( final StringBuilder sb, final String s ) {
+ if ( sb.length() > 0 ) {
+ sb.append( " " );
}
+ sb.append( s.trim() );
}
@Override
- public String getSequenceSymbol() {
- return _symbol;
- }
-
- @Override
- public boolean isEmpty() {
- return ( ForesterUtil.isEmpty( getAccession() ) && ForesterUtil.isEmpty( getSequenceName() )
- && ForesterUtil.isEmpty( getTaxonomyScientificName() )
- && ForesterUtil.isEmpty( getTaxonomyIdentifier() ) && ForesterUtil.isEmpty( getSequenceSymbol() ) );
- }
-
- @Override
- public String getProvider() {
- return _provider;
- }
-
- public void setProvider( final String provider ) {
- _provider = provider;
- }
-
- @Override
- public String getGeneName() {
- return null;
- }
-
- @Override
- public List<GoTerm> getGoTerms() {
- return null;
- }
-
- @Override
- public List<Accession> getCrossReferences() {
+ public MolecularSequence getMolecularSequence() {
+ // TODO Auto-generated method stub
return null;
}
}
git://source.jalview.org / jalview.git / blobdiff