-/**\r
- * \r
- */\r
-package jalview.ws.dbsources;\r
-\r
-import java.io.File;\r
-import java.io.FileReader;\r
-import java.io.IOException;\r
-import java.util.Enumeration;\r
-import java.util.Hashtable;\r
-import java.util.Vector;\r
-\r
-import org.exolab.castor.xml.Unmarshaller;\r
-\r
-import com.stevesoft.pat.Regex;\r
-\r
-import jalview.datamodel.Alignment;\r
-import jalview.datamodel.AlignmentI;\r
-import jalview.datamodel.DBRefEntry;\r
-import jalview.datamodel.DBRefSource;\r
-import jalview.datamodel.PDBEntry;\r
-import jalview.datamodel.SequenceFeature;\r
-import jalview.datamodel.SequenceI;\r
-import jalview.datamodel.UniprotEntry;\r
-import jalview.datamodel.UniprotFile;\r
-import jalview.io.FormatAdapter;\r
-import jalview.io.IdentifyFile;\r
-import jalview.ws.DBRefFetcher;\r
-import jalview.ws.ebi.EBIFetchClient;\r
-import jalview.ws.seqfetcher.DbSourceProxy;\r
-import jalview.ws.seqfetcher.DbSourceProxyImpl;\r
-\r
-/**\r
- * @author JimP\r
- * \r
- */\r
-public class Uniprot extends DbSourceProxyImpl implements DbSourceProxy\r
-{\r
- public Uniprot() {\r
- super();\r
- addDbSourceProperty(DBRefSource.SEQDB, DBRefSource.SEQDB);\r
- addDbSourceProperty(DBRefSource.PROTSEQDB);\r
-// addDbSourceProperty(DBRefSource.MULTIACC, new Integer(50));\r
- }\r
-\r
- /*\r
- * (non-Javadoc)\r
- * \r
- * @see jalview.ws.DbSourceProxy#getAccessionSeparator()\r
- */\r
- public String getAccessionSeparator()\r
- {\r
- return null; // ";";\r
- }\r
-\r
- /*\r
- * (non-Javadoc)\r
- * \r
- * @see jalview.ws.DbSourceProxy#getAccessionValidator()\r
- */\r
- public Regex getAccessionValidator()\r
- {\r
- return null;\r
- }\r
-\r
- /*\r
- * (non-Javadoc)\r
- * \r
- * @see jalview.ws.DbSourceProxy#getDbSource()\r
- */\r
- public String getDbSource()\r
- {\r
- return DBRefSource.UNIPROT;\r
- }\r
-\r
- /*\r
- * (non-Javadoc)\r
- * \r
- * @see jalview.ws.DbSourceProxy#getDbVersion()\r
- */\r
- public String getDbVersion()\r
- {\r
- return "0"; // we really don't know what version we're on.\r
- }\r
-\r
- private EBIFetchClient ebi = null;\r
-\r
- public Vector getUniprotEntries(File file)\r
- {\r
- UniprotFile uni = new UniprotFile();\r
- try\r
- {\r
- // 1. Load the mapping information from the file\r
- org.exolab.castor.mapping.Mapping map = new org.exolab.castor.mapping.Mapping(uni.getClass().getClassLoader());\r
- java.net.URL url = getClass().getResource("/uniprot_mapping.xml");\r
- map.loadMapping(url);\r
-\r
- // 2. Unmarshal the data\r
- Unmarshaller unmar = new Unmarshaller(uni);\r
- unmar.setIgnoreExtraElements(true);\r
- unmar.setMapping(map);\r
-\r
- uni = (UniprotFile) unmar.unmarshal(new FileReader(file));\r
- }\r
- catch (Exception e)\r
- {\r
- System.out.println("Error getUniprotEntries() " + e);\r
- }\r
-\r
- return uni.getUniprotEntries();\r
- }\r
-\r
- /*\r
- * (non-Javadoc)\r
- * \r
- * @see jalview.ws.DbSourceProxy#getSequenceRecords(java.lang.String[])\r
- */\r
- public AlignmentI getSequenceRecords(String queries) throws Exception\r
- {\r
- startQuery();\r
- try\r
- {\r
- Alignment al=null;\r
- ebi = new EBIFetchClient();\r
- StringBuffer result=new StringBuffer();\r
- // uniprotxml parameter required since december 2007\r
- File file = ebi.fetchDataAsFile("uniprot:" + queries, "uniprotxml", null);\r
- Vector entries = getUniprotEntries(file);\r
-\r
- if (entries != null)\r
- {\r
- // First, make the new sequences\r
- Enumeration en = entries.elements();\r
- while (en.hasMoreElements())\r
- {\r
- UniprotEntry entry = (UniprotEntry) en.nextElement();\r
-\r
- StringBuffer name = new StringBuffer(">UniProt/Swiss-Prot");\r
- Enumeration en2 = entry.getAccession().elements();\r
- while (en2.hasMoreElements())\r
- {\r
- name.append("|");\r
- name.append(en2.nextElement());\r
- }\r
- en2 = entry.getName().elements();\r
- while (en2.hasMoreElements())\r
- {\r
- name.append("|");\r
- name.append(en2.nextElement());\r
- }\r
-\r
- if (entry.getProtein()!=null && entry.getProtein().getName()!=null)\r
- {\r
- for (int nm=0,nmSize=entry.getProtein().getName().size(); nm<nmSize;nm++)\r
- {\r
- name.append(" " + entry.getProtein().getName().elementAt(nm));\r
- }\r
- }\r
-\r
- result.append(name + "\n"\r
- + entry.getUniprotSequence().getContent() + "\n");\r
-\r
- }\r
-\r
- // Then read in the features and apply them to the dataset\r
- al = parseResult(result.toString());\r
- if (al!=null)\r
- {\r
- // Decorate the alignment with database entries.\r
- addUniprotXrefs(al, entries);\r
- } else {\r
- results = result;\r
- }\r
- }\r
- stopQuery();\r
- return al;\r
- } catch (Exception e)\r
- {\r
- stopQuery();\r
- throw(e);\r
- }\r
- }\r
-\r
- /**\r
- * add an ordered set of UniprotEntry objects to an ordered set of seuqences.\r
- * \r
- * @param al -\r
- * a sequence of n sequences\r
- * @param entries\r
- * a seuqence of n uniprot entries to be analysed.\r
- */\r
- public void addUniprotXrefs(Alignment al, Vector entries)\r
- {\r
- for (int i = 0; i < entries.size(); i++)\r
- {\r
- UniprotEntry entry = (UniprotEntry) entries.elementAt(i);\r
- Enumeration e = entry.getDbReference().elements();\r
- Vector onlyPdbEntries = new Vector();\r
- Vector dbxrefs = new Vector();\r
- while (e.hasMoreElements())\r
- {\r
- PDBEntry pdb = (PDBEntry) e.nextElement();\r
- if (!pdb.getType().equals("PDB"))\r
- {\r
- DBRefEntry dbr = new DBRefEntry();\r
- dbr.setSource(pdb.getType());\r
- dbr.setAccessionId(pdb.getId());\r
- dbr.setVersion(DBRefSource.UNIPROT+":"+getDbVersion());\r
- dbxrefs.addElement(dbr);\r
- continue;\r
- }\r
-\r
- onlyPdbEntries.addElement(pdb);\r
- }\r
- SequenceI sq = al.getSequenceAt(i);\r
- while (sq.getDatasetSequence()!=null)\r
- {\r
- sq = sq.getDatasetSequence();\r
- }\r
-\r
- Enumeration en2 = entry.getAccession().elements();\r
- while (en2.hasMoreElements())\r
- {\r
- // we always add as uniprot if we retrieved from uniprot or uniprot name\r
- sq.addDBRef(\r
- new DBRefEntry(DBRefSource.UNIPROT, getDbVersion(), en2.nextElement()\r
- .toString()));\r
- }\r
- en2 = dbxrefs.elements();\r
- while (en2.hasMoreElements())\r
- {\r
- // we always add as uniprot if we retrieved from uniprot or uniprot name\r
- sq.addDBRef((DBRefEntry) en2.nextElement());\r
- \r
- }\r
- sq.setPDBId(onlyPdbEntries);\r
- if (entry.getFeature() != null)\r
- {\r
- e = entry.getFeature().elements();\r
- while (e.hasMoreElements())\r
- {\r
- SequenceFeature sf = (SequenceFeature) e.nextElement();\r
- sf.setFeatureGroup("Uniprot");\r
- sq.addSequenceFeature(sf);\r
- }\r
- }\r
- }\r
- }\r
-\r
- /*\r
- * (non-Javadoc)\r
- * \r
- * @see jalview.ws.DbSourceProxy#isValidReference(java.lang.String)\r
- */\r
- public boolean isValidReference(String accession)\r
- {\r
- return true;\r
- }\r
- /**\r
- * return LDHA_CHICK uniprot entry\r
- */\r
- public String getTestQuery()\r
- {\r
- return "P00340";\r
- }\r
- public String getDbName()\r
- {\r
- return "Uniprot"; // getDbSource();\r
- }\r
-}\r
+/**
+ *
+ */
+package jalview.ws.dbsources;
+
+import java.io.File;
+import java.io.FileReader;
+import java.io.IOException;
+import java.util.Enumeration;
+import java.util.Hashtable;
+import java.util.Vector;
+
+import org.exolab.castor.xml.Unmarshaller;
+
+import com.stevesoft.pat.Regex;
+
+import jalview.datamodel.Alignment;
+import jalview.datamodel.AlignmentI;
+import jalview.datamodel.DBRefEntry;
+import jalview.datamodel.DBRefSource;
+import jalview.datamodel.PDBEntry;
+import jalview.datamodel.SequenceFeature;
+import jalview.datamodel.SequenceI;
+import jalview.datamodel.UniprotEntry;
+import jalview.datamodel.UniprotFile;
+import jalview.io.FormatAdapter;
+import jalview.io.IdentifyFile;
+import jalview.ws.DBRefFetcher;
+import jalview.ws.ebi.EBIFetchClient;
+import jalview.ws.seqfetcher.DbSourceProxy;
+import jalview.ws.seqfetcher.DbSourceProxyImpl;
+
+/**
+ * @author JimP
+ *
+ */
+public class Uniprot extends DbSourceProxyImpl implements DbSourceProxy
+{
+ public Uniprot() {
+ super();
+ addDbSourceProperty(DBRefSource.SEQDB, DBRefSource.SEQDB);
+ addDbSourceProperty(DBRefSource.PROTSEQDB);
+// addDbSourceProperty(DBRefSource.MULTIACC, new Integer(50));
+ }
+
+ /*
+ * (non-Javadoc)
+ *
+ * @see jalview.ws.DbSourceProxy#getAccessionSeparator()
+ */
+ public String getAccessionSeparator()
+ {
+ return null; // ";";
+ }
+
+ /*
+ * (non-Javadoc)
+ *
+ * @see jalview.ws.DbSourceProxy#getAccessionValidator()
+ */
+ public Regex getAccessionValidator()
+ {
+ return null;
+ }
+
+ /*
+ * (non-Javadoc)
+ *
+ * @see jalview.ws.DbSourceProxy#getDbSource()
+ */
+ public String getDbSource()
+ {
+ return DBRefSource.UNIPROT;
+ }
+
+ /*
+ * (non-Javadoc)
+ *
+ * @see jalview.ws.DbSourceProxy#getDbVersion()
+ */
+ public String getDbVersion()
+ {
+ return "0"; // we really don't know what version we're on.
+ }
+
+ private EBIFetchClient ebi = null;
+
+ public Vector getUniprotEntries(File file)
+ {
+ UniprotFile uni = new UniprotFile();
+ try
+ {
+ // 1. Load the mapping information from the file
+ org.exolab.castor.mapping.Mapping map = new org.exolab.castor.mapping.Mapping(uni.getClass().getClassLoader());
+ java.net.URL url = getClass().getResource("/uniprot_mapping.xml");
+ map.loadMapping(url);
+
+ // 2. Unmarshal the data
+ Unmarshaller unmar = new Unmarshaller(uni);
+ unmar.setIgnoreExtraElements(true);
+ unmar.setMapping(map);
+
+ uni = (UniprotFile) unmar.unmarshal(new FileReader(file));
+ }
+ catch (Exception e)
+ {
+ System.out.println("Error getUniprotEntries() " + e);
+ }
+
+ return uni.getUniprotEntries();
+ }
+
+ /*
+ * (non-Javadoc)
+ *
+ * @see jalview.ws.DbSourceProxy#getSequenceRecords(java.lang.String[])
+ */
+ public AlignmentI getSequenceRecords(String queries) throws Exception
+ {
+ startQuery();
+ try
+ {
+ Alignment al=null;
+ ebi = new EBIFetchClient();
+ StringBuffer result=new StringBuffer();
+ // uniprotxml parameter required since december 2007
+ File file = ebi.fetchDataAsFile("uniprot:" + queries, "uniprotxml", null);
+ Vector entries = getUniprotEntries(file);
+
+ if (entries != null)
+ {
+ // First, make the new sequences
+ Enumeration en = entries.elements();
+ while (en.hasMoreElements())
+ {
+ UniprotEntry entry = (UniprotEntry) en.nextElement();
+
+ StringBuffer name = new StringBuffer(">UniProt/Swiss-Prot");
+ Enumeration en2 = entry.getAccession().elements();
+ while (en2.hasMoreElements())
+ {
+ name.append("|");
+ name.append(en2.nextElement());
+ }
+ en2 = entry.getName().elements();
+ while (en2.hasMoreElements())
+ {
+ name.append("|");
+ name.append(en2.nextElement());
+ }
+
+ if (entry.getProtein()!=null && entry.getProtein().getName()!=null)
+ {
+ for (int nm=0,nmSize=entry.getProtein().getName().size(); nm<nmSize;nm++)
+ {
+ name.append(" " + entry.getProtein().getName().elementAt(nm));
+ }
+ }
+
+ result.append(name + "\n"
+ + entry.getUniprotSequence().getContent() + "\n");
+
+ }
+
+ // Then read in the features and apply them to the dataset
+ al = parseResult(result.toString());
+ if (al!=null)
+ {
+ // Decorate the alignment with database entries.
+ addUniprotXrefs(al, entries);
+ } else {
+ results = result;
+ }
+ }
+ stopQuery();
+ return al;
+ } catch (Exception e)
+ {
+ stopQuery();
+ throw(e);
+ }
+ }
+
+ /**
+ * add an ordered set of UniprotEntry objects to an ordered set of seuqences.
+ *
+ * @param al -
+ * a sequence of n sequences
+ * @param entries
+ * a seuqence of n uniprot entries to be analysed.
+ */
+ public void addUniprotXrefs(Alignment al, Vector entries)
+ {
+ for (int i = 0; i < entries.size(); i++)
+ {
+ UniprotEntry entry = (UniprotEntry) entries.elementAt(i);
+ Enumeration e = entry.getDbReference().elements();
+ Vector onlyPdbEntries = new Vector();
+ Vector dbxrefs = new Vector();
+ while (e.hasMoreElements())
+ {
+ PDBEntry pdb = (PDBEntry) e.nextElement();
+ DBRefEntry dbr = new DBRefEntry();
+ dbr.setSource(pdb.getType());
+ dbr.setAccessionId(pdb.getId());
+ dbr.setVersion(DBRefSource.UNIPROT+":"+getDbVersion());
+ dbxrefs.addElement(dbr);
+ if (!pdb.getType().equals("PDB"))
+ {
+ continue;
+ }
+
+ onlyPdbEntries.addElement(pdb);
+ }
+ SequenceI sq = al.getSequenceAt(i);
+ while (sq.getDatasetSequence()!=null)
+ {
+ sq = sq.getDatasetSequence();
+ }
+
+ Enumeration en2 = entry.getAccession().elements();
+ while (en2.hasMoreElements())
+ {
+ // we always add as uniprot if we retrieved from uniprot or uniprot name
+ sq.addDBRef(
+ new DBRefEntry(DBRefSource.UNIPROT, getDbVersion(), en2.nextElement()
+ .toString()));
+ }
+ en2 = dbxrefs.elements();
+ while (en2.hasMoreElements())
+ {
+ // we always add as uniprot if we retrieved from uniprot or uniprot name
+ sq.addDBRef((DBRefEntry) en2.nextElement());
+
+ }
+ sq.setPDBId(onlyPdbEntries);
+ if (entry.getFeature() != null)
+ {
+ e = entry.getFeature().elements();
+ while (e.hasMoreElements())
+ {
+ SequenceFeature sf = (SequenceFeature) e.nextElement();
+ sf.setFeatureGroup("Uniprot");
+ sq.addSequenceFeature(sf);
+ }
+ }
+ }
+ }
+
+ /*
+ * (non-Javadoc)
+ *
+ * @see jalview.ws.DbSourceProxy#isValidReference(java.lang.String)
+ */
+ public boolean isValidReference(String accession)
+ {
+ return true;
+ }
+ /**
+ * return LDHA_CHICK uniprot entry
+ */
+ public String getTestQuery()
+ {
+ return "P00340";
+ }
+ public String getDbName()
+ {
+ return "Uniprot"; // getDbSource();
+ }
+}
git://source.jalview.org / jalview.git / commitdiff