/*
* Returns the ProteinProphetFile... this gives us a query-able object which can
* determine if a protein accession was identified by an experiment, as well as
* specific peptides.
*/
private static ProteinProphetFile processProteinProphetFile(String protXMLFileName)
{
const double fdr_threshold = 0.01; // 1% false discovery rate
double fdr = setFDRThreshold(protXMLFileName, fdr_threshold);
List <String> proteinGroupsData = extractProteinGroupsData(protXMLFileName);
List <String> filteredProteinsData = filterProteinsData(proteinGroupsData);
Dictionary <String, List <String> > proteinsToPeptides = extractPeptides(filteredProteinsData);
ProteinProphetFile ppf = new ProteinProphetFile(protXMLFileName, proteinsToPeptides, fdr,
protein_probablity_threshold);
ppf.getProteinProphetResult().SetProteinGroup(ProteinProphetEvaluator.ExtractPositiveProteinGroups(protXMLFileName));
return(ppf);
}
/*
* 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());
}
public static void main(String[] args)
{
String original_protein_prophet_output = "/Users/apell035/workspace/RealTimeMS/data/processed_files/protein_prophet_output/Alex_Mac_MS_QC_60min_interact.prot.xml";
String testFile = "/Users/apell035/workspace/RealTimeMS/data/processed_files/protein_prophet_output/Alex_Mac_MS_QC_240min_interact.prot.xml";
// String rdb_file =
// "data/tsv_files/MS_QC_60min_result_database_NEW_REPROCESSED.tsv";
// ResultDatabase rdb = Loader.parseResultDatabase(rdb_file);
ProteinProphetFile ogppf = processProteinProphetFile(original_protein_prophet_output);
ProteinProphetFile ppf = processProteinProphetFile(testFile);
Dictionary <String, List <String> > setOfUnidentifiedPeptides = ogppf.comparePeptides(ppf);
List <int> unusedSpectra = new List <int>();
// String temp =
// "C:\\Users\\Alexander\\workspace\\tpp\\fromMac\\2018-10-29_explore_Experiment_480_xCorr_3.5_numDB_5_ppmTol_2.0E-6_rtWindow_2.0_comet_search_interact.prot.xml";
// ProteinProphetResult numProtID =
// numProteinsIdentified(original_protein_prophet_output);
// List<String> proteins =
// extractIdentifiedProteinNames(original_protein_prophet_output);
// System.out.println(proteins);
}
/*
* 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());
}
/*
* Extract the number of proteins identified at a given FDR threshold and
* protein probability threshold
*/
public static ProteinProphetResult getProteinProphetResult(String protXMLFileName)
{
ProteinProphetFile ppf = processProteinProphetFile(protXMLFileName);
return(ppf.getProteinProphetResult());
}