Add in previously missed protein aliases...

emckee2006 · emckee2006 · commit 246525ba13fa · 2017-02-08T14:26:46.000-05:00
diff --git a/biojava-core/src/main/java/org/biojava/nbio/core/sequence/loader/UniprotProxySequenceReader.java b/biojava-core/src/main/java/org/biojava/nbio/core/sequence/loader/UniprotProxySequenceReader.java
@@ -363,6 +363,13 @@ public ArrayList<String> getProteinAliases() throws XPathExpressionException {
 		for (Element element : keyWordElementList) {
 			Element fullNameElement = XMLHelper.selectSingleElement(element, "fullName");
 			aliasList.add(fullNameElement.getTextContent());
+			Element shortNameElement = XMLHelper.selectSingleElement(element, "shortName");
+			if(null != shortNameElement) {
+				String shortName = shortNameElement.getTextContent();
+				if(null != shortName && !shortName.trim().isEmpty()) {
+					aliasList.add(shortName);
+				}
+			}
 		}
 		keyWordElementList = XMLHelper.selectElements(proteinElement, "recommendedName");
 		for (Element element : keyWordElementList) {
@@ -376,6 +383,13 @@ public ArrayList<String> getProteinAliases() throws XPathExpressionException {
 				}
 			}
 		}
+		Element cdAntigen = XMLHelper.selectSingleElement(proteinElement, "cdAntigenName");
+		if(null != cdAntigen) {
+			String cdAntigenName = cdAntigen.getTextContent();
+			if(null != cdAntigenName && !cdAntigenName.trim().isEmpty()) {
+				aliasList.add(cdAntigenName);
+			}
+		}
 
 		return aliasList;
 	}
@@ -392,12 +406,13 @@ public ArrayList<String> getGeneAliases() throws XPathExpressionException {
 		}
 		Element uniprotElement = uniprotDoc.getDocumentElement();
 		Element entryElement = XMLHelper.selectSingleElement(uniprotElement, "entry");
-		Element proteinElement = XMLHelper.selectSingleElement(entryElement, "gene");
-		ArrayList<Element> keyWordElementList = XMLHelper.selectElements(proteinElement, "name");
-		for (Element element : keyWordElementList) {
-			aliasList.add(element.getTextContent());
+		ArrayList<Element> proteinElements = XMLHelper.selectElements(entryElement, "gene");
+		for(Element proteinElement : proteinElements) {
+			ArrayList<Element> keyWordElementList = XMLHelper.selectElements(proteinElement, "name");
+			for (Element element : keyWordElementList) {
+				aliasList.add(element.getTextContent());
+			}
 		}
-
 		return aliasList;
 	}
 

Original file line number	Diff line number	Diff line change
`@@ -363,6 +363,13 @@ public ArrayList<String> getProteinAliases() throws XPathExpressionException {`
`363`	`363`	`for (Element element : keyWordElementList) {`
`364`	`364`	`Element fullNameElement = XMLHelper.selectSingleElement(element, "fullName");`
`365`	`365`	`aliasList.add(fullNameElement.getTextContent());`
	`366`	`+ Element shortNameElement = XMLHelper.selectSingleElement(element, "shortName");`
	`367`	`+ if(null != shortNameElement) {`
	`368`	`+ String shortName = shortNameElement.getTextContent();`
	`369`	`+ if(null != shortName && !shortName.trim().isEmpty()) {`
	`370`	`+ aliasList.add(shortName);`
	`371`	`+ }`
	`372`	`+ }`
`366`	`373`	`}`
`367`	`374`	`keyWordElementList = XMLHelper.selectElements(proteinElement, "recommendedName");`
`368`	`375`	`for (Element element : keyWordElementList) {`
`@@ -376,6 +383,13 @@ public ArrayList<String> getProteinAliases() throws XPathExpressionException {`
`376`	`383`	`}`
`377`	`384`	`}`
`378`	`385`	`}`
	`386`	`+ Element cdAntigen = XMLHelper.selectSingleElement(proteinElement, "cdAntigenName");`
	`387`	`+ if(null != cdAntigen) {`
	`388`	`+ String cdAntigenName = cdAntigen.getTextContent();`
	`389`	`+ if(null != cdAntigenName && !cdAntigenName.trim().isEmpty()) {`
	`390`	`+ aliasList.add(cdAntigenName);`
	`391`	`+ }`
	`392`	`+ }`
`379`	`393`
`380`	`394`	`return aliasList;`
`381`	`395`	`}`
`@@ -392,12 +406,13 @@ public ArrayList<String> getGeneAliases() throws XPathExpressionException {`
`392`	`406`	`}`
`393`	`407`	`Element uniprotElement = uniprotDoc.getDocumentElement();`
`394`	`408`	`Element entryElement = XMLHelper.selectSingleElement(uniprotElement, "entry");`
`395`		`- Element proteinElement = XMLHelper.selectSingleElement(entryElement, "gene");`
`396`		`- ArrayList<Element> keyWordElementList = XMLHelper.selectElements(proteinElement, "name");`
`397`		`- for (Element element : keyWordElementList) {`
`398`		`- aliasList.add(element.getTextContent());`
	`409`	`+ ArrayList<Element> proteinElements = XMLHelper.selectElements(entryElement, "gene");`
	`410`	`+ for(Element proteinElement : proteinElements) {`
	`411`	`+ ArrayList<Element> keyWordElementList = XMLHelper.selectElements(proteinElement, "name");`
	`412`	`+ for (Element element : keyWordElementList) {`
	`413`	`+ aliasList.add(element.getTextContent());`
	`414`	`+ }`
`399`	`415`	`}`
`400`		`-`
`401`	`416`	`return aliasList;`
`402`	`417`	`}`
`403`	`418`