private List<String> initIdentifiedSpectrumList() {
// 1. create an empty list for storing identified spectrum ids.
List<String> identifiedList = new ArrayList<String>();
// 2. get all gel free peptide spectrum references
Collection<IndexElement> gelFreeSpectrumRefElements =
indexer.getIndexElements(PrideXmlXpath.GELFREE_PEPTIDE_SPEC_REF.getXpath());
// 3. iterate over each spectrum reference
for (IndexElement spectrumRefElement : gelFreeSpectrumRefElements) {
// get id from index element using pattern
String id = getIDByPattern(spectrumRefElement, TAG_PATTERN, false);
identifiedList.add(id);
}
// 4. get all two dimentional peptide spectrum references
Collection<IndexElement> twoDimSpectrumRefElements =
indexer.getIndexElements(PrideXmlXpath.TWOD_PEPTIDE_SPEC_REF.getXpath());
// 5. iterate over each spectrum reference
for (IndexElement twoDimSpectrumRefElement : twoDimSpectrumRefElements) {
String id = getIDByPattern(twoDimSpectrumRefElement, TAG_PATTERN, false);
identifiedList.add(id);
}
return identifiedList;
}
/**
* Get the number of peptides.
*
* @return int the count of entries
*/
public int getNumberOfPeptides() {
int cnt = 0;
// gel free
Collection<IndexElement> gelFreePeptides =
indexer.getIndexElements(PrideXmlXpath.GELFREE_PEPTIDE.getXpath());
cnt += gelFreePeptides == null ? 0 : gelFreePeptides.size();
// two dim
Collection<IndexElement> twoDimPeptides =
indexer.getIndexElements(PrideXmlXpath.TWOD_PEPTIDE.getXpath());
cnt += twoDimPeptides == null ? 0 : twoDimPeptides.size();
return cnt;
}
/**
* Get all peptide xml strings within an identifiation.
*
* @param identId identificatio id
* @return List<String> a list of peptide xml strings
*/
public List<String> getPeptideXmlStrings(String identId) {
List<String> peptides = new ArrayList<String>();
List<IndexElement> elements = identToPeptideMap.get(identId);
for (IndexElement element : elements) {
peptides.add(indexer.getXmlByIndexElement(element));
}
return peptides;
}
public String getTwoDimIdentXmlString(String id) {
String result = null;
if (twoDimAccMap != null && twoDimAccMap.containsKey(id)) {
result = indexer.getXmlByIndexElement(twoDimAccMap.get(id));
}
return result;
}
public String getGelFreeIdentXmlString(String id) {
String result = null;
if (gelFreeAccMap != null && gelFreeAccMap.containsKey(id)) {
result = indexer.getXmlByIndexElement(gelFreeAccMap.get(id));
}
return result;
}
public String getSpectrumXmlString(String id) {
String result = null;
if (spectrumIdMap != null && spectrumIdMap.containsKey(id)) {
result = indexer.getXmlByIndexElement(spectrumIdMap.get(id));
}
return result;
}
/**
* Get the XML of the first element matched by xpath and within the experiment range.
*
* @param xpath xpath to find the element.
* @return String XML string matched.
*/
private String getFirstXmlString(String xpath) {
String xml = null;
Iterator<String> xmlStrs = indexer.getXmlStringIterator(xpath);
if (xmlStrs != null && xmlStrs.hasNext()) {
xml = xmlStrs.next();
}
return xml;
}
/**
* Get the corresponding xml string for a peptide.
*
* @param identId identification id.
* @param index peptide index within the identification
* @return String xml string
*/
public String getPeptideXmlString(String identId, int index) {
String xml = null;
List<IndexElement> elements = identToPeptideMap.get(identId);
if (elements != null && index >= 0 && index < elements.size()) {
xml = indexer.getXmlByIndexElement(elements.get(index));
}
return xml;
}
public String getExpCollectionVersionString() {
Collection<IndexElement> indexElements =
indexer.getIndexElements(PrideXmlXpath.EXP_COLLECTION.getXpath());
String version = null;
for (IndexElement indexElement : indexElements) {
version = getIDByPattern(indexElement, VERSION_PATTERN, true);
break;
}
return version;
}
private Map<String, List<IndexElement>> initPeptideCacheMap() {
// 1. create a empty identification to peptide mapping map
Map<String, List<IndexElement>> identPeptideMap =
new LinkedHashMap<String, List<IndexElement>>();
// 2. get all gel free peptide index element
Collection<IndexElement> gelFreePeptideElements =
indexer.getIndexElements(PrideXmlXpath.GELFREE_PEPTIDE.getXpath());
// 3. iterate over each gel free peptide index element
for (IndexElement gelFreePeptideElement : gelFreePeptideElements) {
String identId = searchForId(gelFreePeptideElement, gelFreeAccMap);
addIdentPeptide(identId, gelFreePeptideElement, identPeptideMap);
}
// 4. get all two dimensional peptide index element
Collection<IndexElement> twoDimPeptideElements =
indexer.getIndexElements(PrideXmlXpath.TWOD_PEPTIDE.getXpath());
// 5. iterate over each peptide index element
for (IndexElement twoDimPeptideElement : twoDimPeptideElements) {
String identId = searchForId(twoDimPeptideElement, twoDimAccMap);
addIdentPeptide(identId, twoDimPeptideElement, identPeptideMap);
}
return identPeptideMap;
}
/**
* Find the id from xml string based indexElement's range.
*
* @param indexElement index element.
* @param pattern This pattern should only match once.
* @return String id or accession.
*/
private String getIDByPattern(IndexElement indexElement, Pattern pattern, boolean quickMatch) {
String id = null;
// first get the XML string (only the first bit) from which to extract the ID
long start = indexElement.getStart();
long stop = quickMatch ? start + XML_CHAR_INCREMENT : indexElement.getStop();
String xml = indexer.getXmlSnippet(start, stop);
// then apply the regex pattern to find the ID
Matcher m = pattern.matcher(xml);
if (m.find()) {
id = m.group(1);
}
return id;
}
private Map<String, IndexElement> initCacheMap(String xpath, Pattern idPattern) {
// 1. create a empty id map
Map<String, IndexElement> idMap = new LinkedHashMap<String, IndexElement>();
// 2. get index elements from xpath
Collection<IndexElement> indexElements = indexer.getIndexElements(xpath);
for (IndexElement indexElement : indexElements) {
// get id from index element using pattern
String id = getIDByPattern(indexElement, idPattern, true);
if (idMap.containsKey(id)) {
logger.error("Ambiguous ID Exception: " + xpath + " \nID: " + id);
} else {
idMap.put(id, indexElement);
}
}
return idMap;
}
private Map<String, IndexElement> initIdentificationCacheMap(String xpath) {
// 1. create a empty id map
Map<String, IndexElement> idMap = new LinkedHashMap<String, IndexElement>();
// 2. get index elements from xpath
Collection<IndexElement> indexElements = indexer.getIndexElements(xpath);
// 3. generated Id index
for (IndexElement indexElement : indexElements) {
// get id from index element using pattern
if (idMap.containsKey(identificationId + "")) {
logger.error("Ambiguous ID Exception: " + xpath + " \nID: " + identificationId);
} else {
idMap.put(identificationId + "", indexElement);
}
identificationId++;
}
return idMap;
}
/**
* @param xpath
* @return
*/
public Iterator<String> getPrideXmlEntries(String xpath) {
return indexer.getXmlStringIterator(xpath);
}
public List<String> getCvLookupXmlStrings() {
return indexer.getXmlStringList(PrideXmlXpath.MZDATA_CVLOOKUP.getXpath());
}
public List<String> getReferenceXmlStrings() {
return indexer.getXmlStringList(PrideXmlXpath.EXP_REF.getXpath());
}