@Test
public void testKBtoWurcsSparql() throws SparqlException {
List<Structure> structures = Ebean.find(Structure.class).findList();
HashSet<String> resultList = new HashSet<>();
String ct = "";
for (Structure structure : structures) {
if (structure.id >= 7400) {
if (structure.glycanst.startsWith("v--")) {
structure.glycanst = structure.glycanst.replace("v--", "FreeEnd--");
}
if (structure.glycanst.startsWith("FreenEnd")) {
structure.glycanst = structure.glycanst.replace("FreenEnd", "FreeEnd");
}
if (structure.glycanst.startsWith("FreeEnd?")) {
structure.glycanst = structure.glycanst.replace("FreeEnd?", "FreeEnd--?");
}
if (structure.glycanst.startsWith("<Gly") || structure.glycanst.contains("0.0000u")) {
continue;
}
System.out.println(structure.getGlycanst());
BuilderWorkspace workspace = new BuilderWorkspace(new GlycanRendererAWT());
workspace.setNotation("cfg"); // cfgbw | uoxf | uoxfcol | text
GlycanRenderer renderer = workspace.getGlycanRenderer();
org.eurocarbdb.application.glycanbuilder.Glycan glycan =
org.eurocarbdb.application.glycanbuilder.Glycan.fromString(structure.glycanst.trim());
if (glycan != null) {
ct = glycan.toGlycoCTCondensed();
System.out.println("this was the ct: " + ct);
GlycoSequenceToWurcsSelectSparql s = new GlycoSequenceToWurcsSelectSparql("glycoct");
SparqlEntity se = new SparqlEntity();
ct = StringUtils.chomp(ct);
se.setValue(
GlycoSequenceToWurcsSelectSparql.FromSequence,
ct.replaceAll("\n", "\\\\n")
.replaceAll("x\\(", "u\\(")
.replaceAll("\\)x", "\\)u")
.trim());
s.setSparqlEntity(se);
logger.debug(s.getSparql());
Query query =
QueryFactory.create(s.getSparql().replaceAll("null", "").replace("?Sequence", ""));
System.out.println(
"Id "
+ structure.id
+ " Query: "
+ s.getSparql().replaceAll("null", "").replace("?Sequence", ""));
QueryExecution qe =
QueryExecutionFactory.sparqlService("http://test.ts.glytoucan.org/sparql", query);
ResultSet rs = qe.execSelect();
List<SparqlEntity> results = new ArrayList<>();
HashSet<String> resultsList = new HashSet<>();
while (rs.hasNext()) {
QuerySolution row = rs.next();
Iterator<String> columns = row.varNames();
SparqlEntity se2 = new SparqlEntity();
while (columns.hasNext()) {
String column = columns.next();
RDFNode cell = row.get(column);
if (cell.isResource()) {
Resource resource = cell.asResource();
// do something maybe with the OntModel???
if (resource.isLiteral()) se.setValue(column, resource.asLiteral().getString());
else se.setValue(column, resource.toString());
} else if (cell.isLiteral()) {
se.setValue(column, cell.asLiteral().getString());
} else if (cell.isAnon()) {
se.setValue(column, "anon");
} else {
se.setValue(column, cell.toString());
}
}
results.add(se);
}
for (SparqlEntity entity : results) {
// System.out.println("results: " + entity.getValue("PrimaryId"));
resultList.add(structure.id + "\t" + entity.getValue("PrimaryId").toString());
}
}
}
}
PrintWriter writer = null;
try {
writer =
new PrintWriter(
new OutputStreamWriter(new FileOutputStream("/tmp/HashSet.txt"), "UTF-8"));
} catch (UnsupportedEncodingException e) {
e.printStackTrace();
} catch (FileNotFoundException e) {
e.printStackTrace();
}
for (String c : resultList) {
System.out.println(c);
writer.println(c);
}
}
@Test
public void testGlycotoucanCTSearch() {
List<Structure> structures = Ebean.find(Structure.class).findList();
String ct = "";
for (Structure structure : structures) {
if (structure.id >= 7400) {
if (structure.glycanst.startsWith("v--")) {
structure.glycanst = structure.glycanst.replace("v--", "FreeEnd--");
}
if (structure.glycanst.startsWith("FreenEnd")) {
structure.glycanst = structure.glycanst.replace("FreenEnd", "FreeEnd");
}
if (structure.glycanst.startsWith("FreeEnd?")) {
structure.glycanst = structure.glycanst.replace("FreeEnd?", "FreeEnd--?");
}
if (structure.glycanst.startsWith("<Gly") || structure.glycanst.contains("0.0000u")) {
continue;
}
System.out.println(structure.getGlycanst());
BuilderWorkspace workspace = new BuilderWorkspace(new GlycanRendererAWT());
workspace.setNotation("cfg"); // cfgbw | uoxf | uoxfcol | text
GlycanRenderer renderer = workspace.getGlycanRenderer();
org.eurocarbdb.application.glycanbuilder.Glycan glycan =
org.eurocarbdb.application.glycanbuilder.Glycan.fromString(structure.glycanst);
ct = glycan.toGlycoCTCondensed();
// System.out.println(ct);
// }
// }
/*String ct = "RES\\n" +
"1b:a-dgal-HEX-1:5\\n" +
"2s:n-acetyl\\n" +
"3b:b-dgal-HEX-1:5\\n" +
"4b:a-lgal-HEX-1:5|6:d\\n" +
"5b:a-dgal-HEX-1:5\\n" +
"6s:n-acetyl\\n" +
"7b:b-dglc-HEX-1:5\\n" +
"8s:n-acetyl\\n" +
"LIN\\n" +
"1:1d(2+1)2n\\n" +
"2:1o(3+1)3d\\n" +
"3:3o(2+1)4d\\n" +
"4:3o(3+1)5d\\n" +
"5:5d(2+1)6n\\n" +
"6:1o(6+1)7d\\n" +
"7:7d(2+1)8n";
ct = "RES\n" +
"1b:b-dglc-HEX-1:5\n" +
"2s:n-acetyl\n" +
"3b:b-dglc-HEX-1:5\n" +
"4s:n-acetyl\n" +
"5b:b-dman-HEX-1:5\n" +
"6b:a-dman-HEX-1:5\n" +
"7b:a-dman-HEX-1:5\n" +
"8b:a-lgal-HEX-1:5|6:d\n" +
"LIN\n" +
"1:1d(2+1)2n\n" +
"2:1o(4+1)3d\n" +
"3:3d(2+1)4n\n" +
"4:3o(4+1)5d\n" +
"5:5o(3+1)6d\n" +
"6:5o(6+1)7d\n" +
"7:1o(6+1)8d\n" +
"UND\n" +
"UND1:100.0:100.0\n" +
"ParentIDs:1|3|5|6|7|8\n" +
"SubtreeLinkageID1:x(-1+1)x\n" +
"RES\n" +
"9b:b-dglc-HEX-1:5\n" +
"10s:n-acetyl\n" +
"11b:a-lgal-HEX-1:5|6:d\n" +
"12b:b-dgal-HEX-1:5\n" +
"13b:a-dgro-dgal-NON-2:6|1:a|2:keto|3:d\n" +
"14s:n-acetyl\n" +
"LIN\n" +
"8:9d(2+1)10n\n" +
"9:9o(3+1)11d\n" +
"10:9o(4+1)12d\n" +
"11:12o(-1+2)13d\n" +
"12:13d(5+1)14n\n" +
"UND2:100.0:100.0\n" +
"ParentIDs:1|3|5|6|7|8\n" +
"SubtreeLinkageID1:x(-1+1)x\n" +
"RES\n" +
"15b:b-dglc-HEX-1:5\n" +
"16s:n-acetyl\n" +
"17b:a-lgal-HEX-1:5|6:d\n" +
"18b:b-dgal-HEX-1:5\n" +
"LIN\n" +
"13:15d(2+1)16n\n" +
"14:15o(3+1)17d\n" +
"15:15o(4+1)18d\n" +
"UND3:100.0:100.0\n" +
"ParentIDs:1|3|5|6|7|8\n" +
"SubtreeLinkageID1:x(-1+1)x\n" +
"RES\n" +
"19b:b-dglc-HEX-1:5\n" +
"20s:n-acetyl\n" +
"21b:b-dgal-HEX-1:5\n" +
"LIN\n" +
"16:19d(2+1)20n\n" +
"17:19o(4+1)21d";
*/
ct = ct.replaceAll("\n", "\\\\n").replaceAll("x\\(", "u\\(").replaceAll("\\)x", "\\)u");
System.out.println("new ct: " + ct);
String queryString =
"PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>\n"
+ "PREFIX glycan: <http://purl.jp/bio/12/glyco/glycan#>\n"
+ "PREFIX wurcs: <http://www.glycoinfo.org/glyco/owl/wurcs#>\n"
+ "SELECT DISTINCT ?glycan ?c\n"
+ "# FROM <http://rdf.glycoinfo.org/wurcs/0.5.0>\n"
+ "# FROM <http://rdf.glycoinfo.org/wurcs/0.5.0/ms>\n"
+ "WHERE {\n"
+ " ?glycan a \tglycan:glycosequence ;\n"
+ "\tglycan:in_carbohydrate_format glycan:carbohydrate_format_glycoct ;\n"
+ "\tglycan:has_sequence\n"
+ "\t\t?c filter(contains(?c, \"RES\\n1b:b-dglc-HEX-1\")) .\n"
+
// "\t\t?c filter(contains(?c, \"" + ct + "\" )) .\n" +
"\n"
+ " }\n"
+ " ORDER BY ?glycan\n"
+ "limit 10";
System.out.println("String: " + queryString + "\t\tID: " + structure.id);
Query query = QueryFactory.create(queryString);
QueryExecution qExe =
QueryExecutionFactory.sparqlService("http://test.ts.glytoucan.org/sparql", query);
ResultSet results = qExe.execSelect();
ResultSetFormatter.out(System.out, results, query);
}
}
}