public final class UniProtWsTools {
+ private static final boolean ALLOW_TAXONOMY_CODE_HACKS = true; //TODO turn off for final realease!
+
public enum Db {
UNKNOWN, UNIPROT;
}
public final static String BASE_URL = "http://www.uniprot.org/";
-
public final static String BASE_EMBL_DB_URL = "http://www.ebi.ac.uk/Tools/dbfetch/dbfetch/embl/";
private final static String URL_ENC = "UTF-8";
// uniprot/expasy accession number format (6 chars):
// letter digit letter-or-digit letter-or-digit letter-or-digit digit
+ // ?: => no back-reference
+ // \A => begin of String
+ // \Z => end of String
private final static Pattern UNIPROT_AC_PATTERN = Pattern
- .compile( "^.*[^a-zA-Z0-9]?([A-NR-ZOPQ]\\d[A-Z0-9]{3}\\d)[^a-zA-Z0-9]?" );
+ .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]\\d[A-Z0-9]{3}\\d)(?:[^a-zA-Z0-9]|\\Z)" );
private final static boolean DEBUG = false;
private static String encode( final String str ) throws UnsupportedEncodingException {
return null;
}
}
-
-
public static List<UniProtTaxonomy> getTaxonomiesFromCommonName( final String cn, final int max_taxonomies_return )
throws IOException {
throws IOException {
// Hack! Craniata? ..
if ( sn.equals( "Drosophila" ) ) {
- return hack( UniProtTaxonomy.DROSOPHILA_GENUS );
+ return uniProtTaxonomyToList( UniProtTaxonomy.DROSOPHILA_GENUS );
}
else if ( sn.equals( "Xenopus" ) ) {
- return hack( UniProtTaxonomy.XENOPUS_GENUS );
+ return uniProtTaxonomyToList( UniProtTaxonomy.XENOPUS_GENUS );
}
+ // else if ( sn.equals( "Nucleariidae and Fonticula group" ) ) {
+ // return hack( UniProtTaxonomy.NUCLEARIIDAE_AND_FONTICULA );
+ // }
final List<String> result = getTaxonomyStringFromScientificName( sn, max_taxonomies_return );
if ( result.size() > 0 ) {
return parseUniProtTaxonomy( result );
public static List<UniProtTaxonomy> getTaxonomiesFromTaxonomyCode( final String code,
final int max_taxonomies_return )
throws IOException {
- String my_code = new String( code );
- // Hacks!
- if ( my_code.equals( "FUGRU" ) ) {
- my_code = "TAKRU";
- }
- else if ( my_code.equals( "CAP" ) ) {
- return hack( UniProtTaxonomy.CAPITELLA_TELATA_SPECIES );
+ final String my_code = new String( code );
+ if ( ALLOW_TAXONOMY_CODE_HACKS ) {
+ final List<UniProtTaxonomy> l = resolveFakeTaxonomyCodes( max_taxonomies_return, my_code );
+ if ( l != null ) {
+ return l;
+ }
}
final List<String> result = getTaxonomyStringFromTaxonomyCode( my_code, max_taxonomies_return );
if ( result.size() > 0 ) {
return null;
}
+ private static List<UniProtTaxonomy> resolveFakeTaxonomyCodes( final int max_taxonomies_return, final String code )
+ throws IOException {
+ if ( code.equals( "CAP" ) ) {
+ return getTaxonomiesFromId( "283909", max_taxonomies_return );
+ }
+ else if ( code.equals( "FUGRU" ) ) {
+ return getTaxonomiesFromId( "31033", max_taxonomies_return );
+ }
+ else if ( code.equals( "GIALA" ) ) {
+ return getTaxonomiesFromId( "5741", max_taxonomies_return );
+ }
+ else if ( code.equals( "TRIVE" ) ) {
+ return getTaxonomiesFromId( "413071", max_taxonomies_return );
+ }
+ else if ( code.equals( "CAPOWC" ) ) {
+ return getTaxonomiesFromId( "192875", max_taxonomies_return );
+ }
+ else if ( code.equals( "SPHARC" ) ) {
+ return getTaxonomiesFromId( "667725", max_taxonomies_return );
+ }
+ else if ( code.equals( "THETRA" ) ) {
+ return getTaxonomiesFromId( "529818", max_taxonomies_return );
+ }
+ else if ( code.equals( "CHLVUL" ) ) {
+ return getTaxonomiesFromId( "574566", max_taxonomies_return );
+ }
+ else if ( code.equals( "CITCLE" ) ) {
+ return getTaxonomiesFromId( "85681", max_taxonomies_return );
+ }
+ else if ( code.equals( "MYCPOP" ) ) {
+ return getTaxonomiesFromId( "85929", max_taxonomies_return );
+ }
+ else if ( code.equals( "AGABB" ) ) {
+ return getTaxonomiesFromId( "597362", max_taxonomies_return );
+ }
+ else if ( code.equals( "BAUCOM" ) ) {
+ return getTaxonomiesFromId( "430998", max_taxonomies_return );
+ }
+ else if ( code.equals( "DICSQU" ) ) {
+ return getTaxonomiesFromId( "114155", max_taxonomies_return );
+ }
+ else if ( code.equals( "FOMPIN" ) ) {
+ return getTaxonomiesFromId( "40483", max_taxonomies_return );
+ }
+ else if ( code.equals( "HYDMA" ) ) {
+ return getTaxonomiesFromId( "6085", max_taxonomies_return );
+ }
+ else if ( code.equals( "MYCFI" ) ) {
+ return getTaxonomiesFromId( "83344", max_taxonomies_return );
+ }
+ else if ( code.equals( "OIDMAI" ) ) {
+ return getTaxonomiesFromId( "78148", max_taxonomies_return );
+ }
+ else if ( code.equals( "OSTRC" ) ) {
+ return getTaxonomiesFromId( "385169", max_taxonomies_return );
+ }
+ else if ( code.equals( "POSPL" ) ) {
+ return getTaxonomiesFromId( "104341", max_taxonomies_return );
+ }
+ else if ( code.equals( "SAICOM" ) ) {
+ return getTaxonomiesFromId( "5606", max_taxonomies_return );
+ }
+ else if ( code.equals( "SERLA" ) ) {
+ return getTaxonomiesFromId( "85982", max_taxonomies_return );
+ }
+ else if ( code.equals( "SPORO" ) ) {
+ return getTaxonomiesFromId( "40563", max_taxonomies_return );
+ }
+ else if ( code.equals( "ACRALC" ) ) {
+ return getTaxonomiesFromId( "398408", max_taxonomies_return );
+ }
+ else if ( code.equals( "THITER" ) ) {
+ return getTaxonomiesFromId( "35720", max_taxonomies_return );
+ }
+ else if ( code.equals( "MYCTHE" ) ) {
+ return getTaxonomiesFromId( "78579", max_taxonomies_return );
+ }
+ else if ( code.equals( "CONPUT" ) ) {
+ return getTaxonomiesFromId( "80637", max_taxonomies_return );
+ }
+ else if ( code.equals( "WOLCOC" ) ) {
+ return getTaxonomiesFromId( "81056", max_taxonomies_return );
+ }
+ else if ( code.equals( "CLAGRA" ) ) {
+ return getTaxonomiesFromId( "27339", max_taxonomies_return );
+ }
+ else if ( code.equals( "XANPAR" ) ) {
+ return getTaxonomiesFromId( "107463", max_taxonomies_return );
+ }
+ else {
+ return null;
+ }
+ }
+
private static List<String> getTaxonomyStringFromCommonName( final String cn, final int max_lines_to_return )
throws IOException {
return queryUniprot( "taxonomy/?query=common%3a%22" + encode( cn ) + "%22&format=tab", max_lines_to_return );
return queryUniprot( "taxonomy/?query=mnemonic%3a%22" + encode( code ) + "%22&format=tab", max_lines_to_return );
}
- private static List<UniProtTaxonomy> hack( final UniProtTaxonomy tax ) {
+ private static List<UniProtTaxonomy> uniProtTaxonomyToList( final UniProtTaxonomy tax ) {
final List<UniProtTaxonomy> l = new ArrayList<UniProtTaxonomy>();
l.add( tax );
return l;
// Ignore empty lines.
}
else if ( line.startsWith( "Taxon" ) ) {
- //TODO next the check format FIXME
+ final String[] items = line.split( "\t" );
+ if ( !( items[ 1 ].equalsIgnoreCase( "Mnemonic" ) && items[ 2 ].equalsIgnoreCase( "Scientific name" )
+ && items[ 3 ].equalsIgnoreCase( "Common name" ) && items[ 4 ].equalsIgnoreCase( "Synonym" )
+ && items[ 5 ].equalsIgnoreCase( "Other Names" ) && items[ 6 ].equalsIgnoreCase( "Reviewed" )
+ && items[ 7 ].equalsIgnoreCase( "Rank" ) && items[ 8 ].equalsIgnoreCase( "Lineage" ) ) ) {
+ throw new IOException( "Unreconized UniProt Taxonomy format: " + line );
+ }
}
else {
if ( line.split( "\t" ).length > 4 ) {
return taxonomies;
}
-
- public static List<String> queryEmblDb( final String query, int max_lines_to_return ) throws IOException {
- return queryDb( query,
- max_lines_to_return,
- BASE_EMBL_DB_URL ) ;
+ public static List<String> queryEmblDb( final String query, final int max_lines_to_return ) throws IOException {
+ return queryDb( query, max_lines_to_return, BASE_EMBL_DB_URL );
}
-
-
-
- public static List<String> queryUniprot( final String query, int max_lines_to_return ) throws IOException {
- return queryDb( query,
- max_lines_to_return,
- BASE_URL ) ;
-
-
+
+ public static List<String> queryUniprot( final String query, final int max_lines_to_return ) throws IOException {
+ return queryDb( query, max_lines_to_return, BASE_URL );
}
- public static List<String> queryDb( final String query,
- int max_lines_to_return,
- final String base_url ) throws IOException {
+ public static List<String> queryDb( final String query, int max_lines_to_return, final String base_url )
+ throws IOException {
if ( ForesterUtil.isEmpty( query ) ) {
throw new IllegalArgumentException( "illegal attempt to use empty query " );
}
String line;
final List<String> result = new ArrayList<String>();
while ( ( line = in.readLine() ) != null ) {
+ if ( DEBUG ) {
+ System.out.println( line );
+ }
result.add( line );
if ( result.size() > max_lines_to_return ) {
break;
in.close();
return result;
}
-
-
+
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 SequenceDatabaseEntry obtainEmblEntry( String query, int max_lines_to_return ) throws IOException {
- final List<String> lines = queryEmblDb( "query", max_lines_to_return );
+ public static SequenceDatabaseEntry obtainEmblEntry( final String query, final int max_lines_to_return )
+ throws IOException {
+ final List<String> lines = queryEmblDb( query, max_lines_to_return );
return EbiDbEntry.createInstanceFromPlainText( lines );
}
}
git://source.jalview.org / jalview.git / blobdiff