/**
* this is what a Mapper does
*
* @param keyin
* @param valuein
* @return iterator over mapped key values
*/
@Nonnull
@Override
public Iterable<KeyValueObject<String, String>> mapValues(
@Nonnull final String annotation, @Nonnull final String sequence) {
List<KeyValueObject<String, String>> holder = new ArrayList<KeyValueObject<String, String>>();
IPeptideDigester digester = getDigester();
IProtein prot = Protein.getProtein(annotation, annotation, sequence, null);
// do a boolean for a peptide belonging to a decoy protein, but use the public isDecoy
// boolean/method in Protein class
boolean isDecoy = prot.isDecoy();
IPolypeptide[] pps = digester.digest(prot);
PeptideModification[] modifications1 = getModifications();
for (int i = 0; i < pps.length; i++) {
IPolypeptide pp = pps[i];
if (!pp.isValid()) continue;
// hadoop write intermediate seq finder
writePeptide(pp, holder);
// if(isDecoy)
// continue; // skip the rest of the loop
// if it is decoy, don't add modifications to it
if (!isDecoy || isGenerateDecoysForModifiedPeptides()) {
// generate modified peptides and add to the output
IModifiedPeptide[] modifications =
ModifiedPolypeptide.buildModifications(pp, modifications1);
for (int m = 0; m < modifications.length; m++) {
IModifiedPeptide modification = modifications[m];
writePeptide(modification, holder);
}
}
}
boolean semiTryptic = digester.isSemiTryptic();
if (semiTryptic) {
IPolypeptide[] semipps = digester.addSemiCleavages(prot);
for (int j = 0; j < semipps.length; j++) {
IPolypeptide semipp = semipps[j];
if (!semipp.isValid()) continue;
writePeptide(semipp, holder);
IModifiedPeptide[] modifications =
ModifiedPolypeptide.buildModifications(semipp, modifications1);
for (int k = 0; k < modifications.length; k++) {
IModifiedPeptide modification = modifications[k];
writePeptide(modification, holder);
}
}
}
return holder;
}
public static IProtein processProtein(final String line) {
String peptide = XMLUtil.extractAttribute(line, "protein_descr");
if (peptide == null) throw new IllegalArgumentException("bad line " + line);
return Protein.getProtein(peptide, "", "", null);
}