Update UniprotProxySequenceReader.java

emckee2006 · emckee2006 · commit 1163074656dc · 2019-04-12T03:43:15.000-04:00
2 fixes:
1. Full Uniprot downloads contain spaces in some sequences, which cause a crash in setContents because space isn't a valid compound. I added a replaceAll to strip out the whitespace.
2. Apparently a shortName element can be repeated. In both cases where we can encounter a shortName element, get the list (perhaps of length 0 or 1) and iterate over each shortName instead of just grabbing the first one.
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
@@ -114,7 +114,7 @@ public UniprotProxySequenceReader(String accession, CompoundSet<C> compoundSet)
 	public UniprotProxySequenceReader(Document document, CompoundSet<C> compoundSet) throws CompoundNotFoundException {
 		setCompoundSet(compoundSet);
 		uniprotDoc = document;
-		String seq = this.getSequence(uniprotDoc);
+		String seq = this.getSequence(uniprotDoc).replaceAll("\\s","");
 		setContents(seq);
 	}
 	/**
@@ -385,23 +385,27 @@ 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);
+			ArrayList<Element> shortNameElements = XMLHelper.selectElements(element, "shortName");
+			for(Element shortNameElement : shortNameElements) {
+				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) {
 			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);
+			ArrayList<Element> shortNameElements = XMLHelper.selectElements(element, "shortName");
+			for(Element shortNameElement : shortNameElements) {
+				if(null != shortNameElement) {
+					String shortName = shortNameElement.getTextContent();
+					if(null != shortName && !shortName.trim().isEmpty()) {
+						aliasList.add(shortName);
+					}
 				}
 			}
 		}
