/*
* Extracts the protein names from proteins not identified with high confidence.
* The fdr_threshold should be a high value (0.2 or higher) for this to be true.
* Extracts the proteins identified above this fdr threshold.
*/
public static List <String> extractNegativeTrainingSetProteinNames(String proteinProphetFile,
double pr_threshold)
{
// setting to an fdr of 0.2 didn't work, because the largest fdr is 0.173...
// double fdr = setFDRThreshold(proteinProphetFile, fdr_threshold);
protein_probablity_threshold = pr_threshold;
List <String> proteinGroupsData = extractNegativeProteinGroupsData(proteinProphetFile);
List <String> filteredProteinsData = filterNegativeTrainingSetProteinData(proteinGroupsData);
Dictionary <String, List <String> > proteinsToPeptides = extractPeptides(filteredProteinsData);
ProteinProphetFile ppf = new ProteinProphetFile(proteinProphetFile, proteinsToPeptides, 1,
protein_probablity_threshold);
return(ppf.getProteinNames());
}
/*
* Returns the list of proteins identified from a protein prophet experiment
*/
public static List <String> extractIdentifiedProteinNames(String protXMLFileName)
{
ProteinProphetFile ppf = processProteinProphetFile(protXMLFileName);
return(ppf.getProteinNames());
}