private String getSubstrateAccession(
Matcher patternSubstrateAccession, SubstrateDatabaseEntry subtratedatabase) {
String accession = null;
if (patternSubstrateAccession.find()) {
accession = patternSubstrateAccession.group(1);
accession = accession.trim();
subtratedatabase.setS_UniprotID(accession);
// System.out.println(accession);
} else {
accession = "n.d.";
subtratedatabase.setS_UniprotID(accession);
// System.out.println(accession);
}
return accession;
}
private String getSubstrateNameSymbolId(
Matcher patternSubstrateName,
SubstrateDatabaseEntry substratedatabase,
CsDatabaseEntry csdatabase,
String entry,
String substrateTaxon)
throws IOException {
String commentS = null;
Matcher patternSubstrateAccession =
getPatternmatcher(
"UniProt\\s+Accession:</th>[^<]+<td><a\\s+href\\s+=\\s+\"[^\"]+\"\\s+target=\"[^\"]+\">"
+ "([^<]+)",
entry);
String accession = getSubstrateAccession(patternSubstrateAccession, substratedatabase);
Matcher patternSubstrateSymbol =
getPatternmatcher(
"Substrate[^<]+</th>[^<]+<td>[^<]+<table>[^<]+<tr>[^<]+<th\\s+class=\"th3\">Definition:</th>[^<]+<td><b>"
+ "[^<]+"
+ "</b></td>[^<]+<tr>[^<]+<th\\s+class=\"th3\">Symbol:</th>"
+ "([^<]+<td><b>)?([^<]+)",
entry);
String symbol = getSubstrateSymbol(patternSubstrateSymbol, substratedatabase);
if (patternSubstrateName.find()) {
String Substratename = "to check";
String Substratesymbol = "to check";
String Substrateaccession = "to check";
substratedatabase.setS_NL_Name(Substratename);
substratedatabase.setS_Name(Substratename);
substratedatabase.setS_Symbol(Substratesymbol);
substratedatabase.setS_UniprotID(Substrateaccession);
Substratename = patternSubstrateName.group(1);
Substratename = Substratename.trim();
Substratename = Substratename.replaceAll(",", "");
Substratename = Substratename.replaceAll(";", "");
substratedatabase.setS_NL_Name(Substratename);
commentS = "Check Substrate Symbol and Accession; add to Substrate Librairy";
BufferedReader bReader = null;
if (substrateTaxon.contains("H**o")) {
bReader =
createBufferedreader(
"/Users/julieklein/Dropbox/ProteasiX/LIBRAIRIES/SubstrateHSALibrairy.txt");
} else if (substrateTaxon.contains("Mus")) {
bReader =
createBufferedreader(
"/Users/julieklein/Dropbox/ProteasiX/LIBRAIRIES/SubstrateMMULibrairy.txt");
} else if (substrateTaxon.contains("Rattus")) {
bReader =
createBufferedreader(
"/Users/julieklein/Dropbox/ProteasiX/LIBRAIRIES/SubstrateRNOLibrairy.txt");
}
String line;
while ((line = bReader.readLine()) != null) {
String splitarray[] = line.split("\t");
String naturallanguage = splitarray[1];
naturallanguage = naturallanguage.replaceAll("\"", "");
naturallanguage = naturallanguage.replaceAll(",", "");
naturallanguage = naturallanguage.replaceAll(";", "");
if (naturallanguage.equalsIgnoreCase(Substratename)) {
Substratesymbol = splitarray[0];
Substratesymbol = Substratesymbol.replaceAll("sept-0", "SEPT");
Substrateaccession = splitarray[2];
if (Substrateaccession.contains("n.d.")) {
substratedatabase.setS_Name("n.d.");
substratedatabase.setS_UniprotID(Substrateaccession);
substratedatabase.setS_Symbol(Substratesymbol);
} else {
String UniprotURL = "http://www.uniprot.org/uniprot/" + Substrateaccession + ".xml";
NodeList entries = getEntries("/uniprot/entry", parseUniprot(UniprotURL));
for (int i = 0; i < entries.getLength(); i++) {
getUniSubstratepproteinname(entries, i, substratedatabase);
String genename = getUniSubstrategenename(entries, i, substratedatabase);
}
// System.out.println(Substrateaccession);
substratedatabase.setS_UniprotID(Substrateaccession);
commentS = "-";
System.out.println(commentS);
}
}
csdatabase.setSubstrate(substratedatabase);
}
} else if (!symbol.contains("n.d.")) {
String Substratename = "to check";
String Substratesymbol = "to check";
String Substrateaccession = "to check";
substratedatabase.setS_NL_Name(Substratename);
substratedatabase.setS_Name(Substratename);
substratedatabase.setS_Symbol(Substratesymbol);
substratedatabase.setS_UniprotID(Substrateaccession);
BufferedReader bReader = null;
if (substrateTaxon.contains("H**o")) {
bReader =
createBufferedreader(
"/Users/julieklein/Dropbox/ProteasiX/LIBRAIRIES/SubstrateHSALibrairy.txt");
} else if (substrateTaxon.contains("Mus")) {
bReader =
createBufferedreader(
"/Users/julieklein/Dropbox/ProteasiX/LIBRAIRIES/SubstrateMMULibrairy.txt");
} else if (substrateTaxon.contains("Rattus")) {
bReader =
createBufferedreader(
"/Users/julieklein/Dropbox/ProteasiX/LIBRAIRIES/SubstrateRNOLibrairy.txt");
}
String line;
while ((line = bReader.readLine()) != null) {
String splitarray[] = line.split("\t");
String librairisymbol = splitarray[0];
librairisymbol = librairisymbol.replaceAll("\"", "");
librairisymbol = librairisymbol.replaceAll("sept-0", "SEPT");
if (librairisymbol.equals(symbol)) {
Substrateaccession = splitarray[2];
if (Substrateaccession.contains("n.d.")) {
substratedatabase.setS_Name("n.d.");
substratedatabase.setS_UniprotID(Substrateaccession);
substratedatabase.setS_Symbol(symbol);
} else {
String UniprotURL = "http://www.uniprot.org/uniprot/" + Substrateaccession + ".xml";
NodeList entries = getEntries("/uniprot/entry", parseUniprot(UniprotURL));
for (int i = 0; i < entries.getLength(); i++) {
getUniSubstratepproteinname(entries, i, substratedatabase);
String genename = getUniSubstrategenename(entries, i, substratedatabase);
}
// System.out.println(Substrateaccession);
substratedatabase.setS_UniprotID(Substrateaccession);
commentS = "-";
System.out.println(commentS);
}
}
csdatabase.setSubstrate(substratedatabase);
}
} else {
String Substratename = "n.d.";
String Substratesymbol = "n.d.";
String Substrateaccession = "n.d";
substratedatabase.setS_NL_Name(Substratename);
substratedatabase.setS_Name(Substratename);
substratedatabase.setS_Symbol(Substratesymbol);
substratedatabase.setS_UniprotID(Substrateaccession);
System.out.println(Substratename);
System.out.println(Substratesymbol);
System.out.println(Substrateaccession);
commentS = "-";
System.out.println(commentS);
csdatabase.setSubstrate(substratedatabase);
}
return commentS;
}