/**
* The Uniprot mappings to other database identifiers for this sequence
*
* @return
* @throws Exception
*/
public LinkedHashMap<String, ArrayList<DBReferenceInfo>> getDatabaseReferences()
throws Exception {
LinkedHashMap<String, ArrayList<DBReferenceInfo>> databaseReferencesHashMap =
new LinkedHashMap<String, ArrayList<DBReferenceInfo>>();
if (uniprotDoc == null) {
return databaseReferencesHashMap;
}
Element uniprotElement = uniprotDoc.getDocumentElement();
Element entryElement = XMLHelper.selectSingleElement(uniprotElement, "entry");
ArrayList<Element> dbreferenceElementList =
XMLHelper.selectElements(entryElement, "dbReference");
for (Element element : dbreferenceElementList) {
String type = element.getAttribute("type");
String id = element.getAttribute("id");
ArrayList<DBReferenceInfo> idlist = databaseReferencesHashMap.get(type);
if (idlist == null) {
idlist = new ArrayList<DBReferenceInfo>();
databaseReferencesHashMap.put(type, idlist);
}
DBReferenceInfo dbreferenceInfo = new DBReferenceInfo(type, id);
ArrayList<Element> propertyElementList = XMLHelper.selectElements(element, "property");
for (Element propertyElement : propertyElementList) {
String propertyType = propertyElement.getAttribute("type");
String propertyValue = propertyElement.getAttribute("value");
dbreferenceInfo.addProperty(propertyType, propertyValue);
}
idlist.add(dbreferenceInfo);
}
return databaseReferencesHashMap;
}
/**
* @param uniprotDoc
* @return
* @throws Exception
*/
private String getSequence(Document uniprotDoc) throws Exception {
Element uniprotElement = uniprotDoc.getDocumentElement();
Element entryElement = XMLHelper.selectSingleElement(uniprotElement, "entry");
Element sequenceElement = XMLHelper.selectSingleElement(entryElement, "sequence");
String seqdata = sequenceElement.getTextContent();
return seqdata;
}
/**
* Pull uniprot key words which is a mixed bag of words associated with this sequence
*
* @return
* @throws Exception
*/
public ArrayList<String> getKeyWords() throws Exception {
ArrayList<String> keyWordsList = new ArrayList<String>();
if (uniprotDoc == null) {
return keyWordsList;
}
Element uniprotElement = uniprotDoc.getDocumentElement();
Element entryElement = XMLHelper.selectSingleElement(uniprotElement, "entry");
ArrayList<Element> keyWordElementList = XMLHelper.selectElements(entryElement, "keyword");
for (Element element : keyWordElementList) {
keyWordsList.add(element.getTextContent());
}
return keyWordsList;
}
/** @return */
public AccessionID getAccession() {
AccessionID accessionID = new AccessionID();
if (uniprotDoc == null) {
return accessionID;
}
try {
Element uniprotElement = uniprotDoc.getDocumentElement();
Element entryElement = XMLHelper.selectSingleElement(uniprotElement, "entry");
Element nameElement = XMLHelper.selectSingleElement(entryElement, "name");
accessionID = new AccessionID(nameElement.getTextContent(), DataSource.UNIPROT);
} catch (Exception e) {
logger.error("Exeception: ", e);
}
return accessionID;
}
/**
* Get the organism name assigned to this sequence
*
* @return
* @throws Exception
*/
public String getOrganismName() throws Exception {
if (uniprotDoc == null) {
return "";
}
Element uniprotElement = uniprotDoc.getDocumentElement();
Element entryElement = XMLHelper.selectSingleElement(uniprotElement, "entry");
Element organismElement = XMLHelper.selectSingleElement(entryElement, "organism");
if (organismElement == null) {
return "";
}
Element nameElement = XMLHelper.selectSingleElement(organismElement, "name");
if (nameElement == null) {
return "";
}
return nameElement.getTextContent();
}
/**
* @param accession
* @return
* @throws Exception
*/
private Document getUniprotXML(String accession) throws Exception {
int index = accession.lastIndexOf(".");
String key = accession;
if (index != -1) {
key = accession.substring(0, index);
}
StringBuilder sb = new StringBuilder();
File f = null;
if (uniprotDirectoryCache != null && uniprotDirectoryCache.length() > 0) {
f = new File(uniprotDirectoryCache + File.separatorChar + key + ".xml");
if (f.exists()) {
FileReader fr = new FileReader(f);
int size = (int) f.length();
char[] data = new char[size];
fr.read(data);
fr.close();
sb.append(data);
index = sb.indexOf("xmlns="); // strip out name space stuff to make it easier on xpath
if (index != -1) {
int lastIndex = sb.indexOf(">", index);
sb.replace(index, lastIndex, "");
}
}
}
// http://www.uniprot.org/uniprot/?query=SORBIDRAFT_03g027040&format=xml
if (sb.length() == 0) {
String uniprotURL = getUniprotbaseURL() + "/uniprot/?query=" + key + "&format=xml";
// String uniprotURL = getUniprotbaseURL() + "/uniprot/" + key + ".xml";
logger.info("Loading: {}", uniprotURL);
URL uniprot = new URL(uniprotURL);
URLConnection uniprotConnection = uniprot.openConnection();
uniprotConnection.setRequestProperty("User-Agent", "BioJava");
BufferedReader in =
new BufferedReader(new InputStreamReader(uniprotConnection.getInputStream()));
String inputLine;
while ((inputLine = in.readLine()) != null) {
sb.append(inputLine);
}
in.close();
index = sb.indexOf("xmlns="); // strip out name space stuff to make it easier on xpath
if (index != -1) {
int lastIndex = sb.indexOf(">", index);
sb.replace(index, lastIndex, "");
}
if (f != null) {
FileWriter fw = new FileWriter(f);
fw.write(sb.toString());
fw.close();
}
}
logger.info("Load complete");
try {
// logger.debug(sb.toString());
Document document =
XMLHelper.inputStreamToDocument(new ByteArrayInputStream(sb.toString().getBytes()));
return document;
} catch (Exception e) {
logger.error("Exception on xml parse of: {}", sb.toString());
}
return null;
}