public void TestMSGFFilter()
{
const string workDir = @"F:\My Documents\Projects\JoshAldrich\AScore\AScore_DLL\TestCase";
var fht = Path.Combine(workDir, "GmaxP_itraq_NiNTA_15_29Apr10_Hawk_03-10-09p_fht.txt");
var param = Path.Combine(workDir, "AScore_CID_0.5Da_ETD_0.5Da_HCD_0.05Da_MSGF1E-12.xml");
var dta = Path.Combine(workDir, "GmaxP_itraq_NiNTA_15_29Apr10_Hawk_03-10-09p_dta.txt");
var resultsFile = Path.Combine(workDir, "GmaxP_itraq_NiNTA_15_29Apr10_Hawk_03-10-09p_fht_ascore.txt");
var peptideMassCalculator = GetDefaultPeptideMassCalculator();
PsmResultsManager dfht = new SequestFHT(fht);
var pfile = new ParameterFileManager(param);
var spectraCache = new SpectraManagerCache(peptideMassCalculator);
spectraCache.OpenFile(dta);
var sw = System.Diagnostics.Stopwatch.StartNew();
var ascoreEngine = new AScore_DLL.AScoreProcessor();
ascoreEngine.RunAScoreOnSingleFile(spectraCache, dfht, pfile, resultsFile);
sw.Stop();
Console.WriteLine("Time elapsed: {0:F1} seconds", sw.Elapsed.TotalSeconds);
Console.WriteLine("Results in " + resultsFile);
}
public void TestMSGFPlusResults()
{
const string workDir = @"F:\My Documents\Projects\JoshAldrich\AScore\AScore_DLL\TestCase";
var fht = Path.Combine(workDir, "CPTAC_CompREF_00_iTRAQ_NiNTA_01b_22Mar12_Lynx_12-02-29_msgfdb_fht.txt");
var param = Path.Combine(workDir, "AScore_CID_0.5Da_ETD_0.5Da_HCD_0.05Da_MSGF1E-12.xml");
var dta = Path.Combine(workDir, "CPTAC_CompREF_00_iTRAQ_NiNTA_01b_22Mar12_Lynx_12-02-29_dta.txt");
var resultsFile = Path.Combine(workDir, "CPTAC_CompREF_00_iTRAQ_NiNTA_01b_22Mar12_Lynx_12-02-29_msgfdb_fht_ascore.txt");
var peptideMassCalculator = GetDefaultPeptideMassCalculator();
PsmResultsManager dfht = new MsgfdbFHT(fht);
var pfile = new ParameterFileManager(param);
var spectraCache = new SpectraManagerCache(peptideMassCalculator);
spectraCache.OpenFile(dta);
var sw = System.Diagnostics.Stopwatch.StartNew();
var ascoreEngine = new AScore_DLL.AScoreProcessor();
ascoreEngine.RunAScoreOnSingleFile(spectraCache, dfht, pfile, resultsFile);
sw.Stop();
Console.WriteLine("Time elapsed: {0:F1} seconds", sw.Elapsed.TotalSeconds);
Console.WriteLine("Results in " + resultsFile);
}
public void EcoliTest2()
{
var datasetname = new string[] {
@"C:\Documents and Settings\aldr699\My Documents2011\SisiTopDown\E_coli_BW_70_bottom_up_23Sep11_Draco_11-07-12_fht.txt"
};
var dtaname = new string[] {
@"C:\Documents and Settings\aldr699\My Documents2011\SisiTopDown\E_coli_BW_70_bottom_up_23Sep11_Draco_11-07-12_dta.txt"
};
var ascParam = new string[] {
@"C:\Documents and Settings\aldr699\My Documents2011\SisiTopDown\parameterFileForMus.xml"
};
var peptideMassCalculator = GetDefaultPeptideMassCalculator();
for (var i = 0; i < dtaname.Length; i++)
{
var outFile = datasetname[i].Substring(0, datasetname[i].Length - 8) + "_AScoreTry.txt";
PsmResultsManager datasetMan = new SequestFHT(datasetname[i]);
//DtaManager dtaManager = new DtaManager(dtaname[i]);
var spectraCache = new SpectraManagerCache(peptideMassCalculator);
spectraCache.OpenFile(dtaname[i]);
var paramFile = new ParameterFileManager(ascParam[i]);
var ascoreEngine = new AScore_DLL.AScoreProcessor();
//ascoreEngine.AlgorithmRun(dtaManager, datasetMan, paramFile, outFile);
ascoreEngine.RunAScoreOnSingleFile(spectraCache, datasetMan, paramFile, outFile);
}
}
public void FengTest2()
{
var datasetname = new string[] {
@"C:\Documents and Settings\aldr699\My Documents2011\FengTest\ETD\B_Mus_con_15Oct10_Hawk_03-10-10p_fht.txt"
};
var dtaname = new string[] {
@"C:\Documents and Settings\aldr699\My Documents2011\FengTest\ETD\B_Mus_con_15Oct10_Hawk_03-10-10p_dta.txt"
};
var ascParam = new string[] {
@"C:\Documents and Settings\aldr699\My Documents2011\FengTest\parameterFileForMusETD2.xml"
};
var peptideMassCalculator = GetDefaultPeptideMassCalculator();
for (var i = 0; i < dtaname.Length; i++)
{
var outFile = datasetname[i].Substring(0, datasetname[i].Length - 8) + "_AScoreTry.txt";
PsmResultsManager datasetMan = new SequestFHT(datasetname[i]);
//DtaManager dtaManager = new DtaManager(dtaname[i]);
var spectraCache = new SpectraManagerCache(peptideMassCalculator);
spectraCache.OpenFile(dtaname[i]);
var paramFile = new ParameterFileManager(ascParam[i]);
var ascoreEngine = new AScore_DLL.AScoreProcessor();
//ascoreEngine.AlgorithmRun(dtaManager, datasetMan, paramFile, outFile);
ascoreEngine.RunAScoreOnSingleFile(spectraCache, datasetMan, paramFile, outFile);
}
}
public void NewTest()
{
var ascore = new string[] { "SOSM_May_P1_R2_13Jun11_Hawk_11-04-02p_fht.txt" };
//"SOSM_May_G2_R2_13Jun11_Hawk_11-04-02p_fht.txt",
//"SOSM_May_M_R2_6Jun11_Hawk_11-04-02p_fht.txt",
//"SOSM_May_P1_R1_6Jun11_Hawk_11-04-02p_fht.txt",
//"SOSM_May_P1_R2_6Jun11_Hawk_11-04-02p_fht.txt"};
const string fhtPath = @"C:\Documents and Settings\aldr699\My Documents2011\SOSM\CID\ForAScore";
const string dtapath = @"C:\Documents and Settings\aldr699\My Documents2011\SOSM";
const string parFile = @"C:\Documents and Settings\aldr699\My Documents2011\SOSM\parameterFile.xml";
var peptideMassCalculator = GetDefaultPeptideMassCalculator();
foreach (var a in ascore)
{
var myFht = Path.Combine(fhtPath, a);
var myDta = Path.Combine(dtapath, a.Substring(0, a.Length - 8) + "_dta.txt");
var outFile = Path.Combine(fhtPath, a.Substring(0, a.Length - 8) + "_AScore.txt");
PsmResultsManager datasetMan = new SequestFHT(myFht);
//DtaManager dtaManager = new DtaManager(myDta);
var spectraCache = new SpectraManagerCache(peptideMassCalculator);
spectraCache.OpenFile(myDta);
var paramFile = new ParameterFileManager(parFile);
var ascoreEngine = new AScore_DLL.AScoreProcessor();
//ascoreEngine.AlgorithmRun(dtaManager, datasetMan, paramFile, outFile);
ascoreEngine.RunAScoreOnSingleFile(spectraCache, datasetMan, paramFile, outFile);
}
}
public void Sisi()
{
const string ascoreP = "parameterFile.xml";
const string direct = @"C:\Users\aldr699\Documents\2012\Sisi_ASCORE\Group2";
var fhtFiles = new List <string> {
"NMR_HetR_UG_01_20Jul11_Andromeda_11-06-19_fhtf.txt",
"NMR_HetR_UG_02_20Jul11_Andromeda_11-06-29_fhtf.txt",
"NMR_HetR_UG_03_20Jul11_Andromeda_11-06-19_fhtf.txt",
"NMR_HetR_LG_01_20Jul11_Andromeda_11-06-29_fhtf.txt",
"NMR_HetR_LG_02_20Jul11_Andromeda_11-06-29_fhtf.txt",
"NMR_HetR_LG_03_20Jul11_Andromeda_11-06-19_fhtf.txt",
"NMR_LaR80a_01_20Jul11_Andromeda_11-02-54_fhtf.txt",
"NMR_LaR80a_02_20Jul11_Andromeda_11-02-56_fhtf.txt",
"NMR_LaR80a_03_20Jul11_Andromeda_11-02-54_fhtf.txt"
};
var dtaFiles = new List <string> {
"NMR_HetR_UG_01_20Jul11_Andromeda_11-06-19_dta.txt",
"NMR_HetR_UG_02_20Jul11_Andromeda_11-06-29_dta.txt",
"NMR_HetR_UG_03_20Jul11_Andromeda_11-06-19_dta.txt",
"NMR_HetR_LG_01_20Jul11_Andromeda_11-06-29_dta.txt",
"NMR_HetR_LG_02_20Jul11_Andromeda_11-06-29_dta.txt",
"NMR_HetR_LG_03_20Jul11_Andromeda_11-06-19_dta.txt",
"NMR_LaR80a_01_20Jul11_Andromeda_11-02-54_dta.txt",
"NMR_LaR80a_02_20Jul11_Andromeda_11-02-56_dta.txt",
"NMR_LaR80a_03_20Jul11_Andromeda_11-02-54_dta.txt"
};
var peptideMassCalculator = GetDefaultPeptideMassCalculator();
for (var i = 0; i < fhtFiles.Count; i++)
{
var tempfht = Path.Combine(direct, fhtFiles[i]);
var tempdta = Path.Combine(direct, dtaFiles[i]);
var ascP = Path.Combine(direct, ascoreP);
var tempout = Path.Combine(direct, fhtFiles[i].Substring(0, fhtFiles[i].Length - 4) + "_AScore2.txt");
PsmResultsManager dsman = new SequestFHT(tempfht);
//DtaManager dtman = new DtaManager(tempdta);
var spectraCache = new SpectraManagerCache(peptideMassCalculator);
spectraCache.OpenFile(tempdta);
var pman = new ParameterFileManager(ascP);
var ascoreEngine = new AScore_DLL.AScoreProcessor();
//ascoreEngine.AlgorithmRun(dtman, dsman, pman, tempout);
ascoreEngine.RunAScoreOnSingleFile(spectraCache, dsman, pman, tempout);
}
}
public void SisiConfirm()
{
var dtaname = new string[] {
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\09302010_MG1655_phospho_S_08_B_100930113735_dta.txt",
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\10012010_MG1655_phospho_S_09_B_101001105913_dta.txt",
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\10012010_MG1655_phospho_s06_101002094806_dta.txt",
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\10012010_MG1655_phospho_s10_101001105913_dta.txt",
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\10022010_MG1655_phospho_S11_rerun_101005165350_dta.txt",
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\10022010_MG1655_phospho_s12_101002094806_dta.txt",
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\10022010_MG1655_phospho_s7_101002094806_dta.txt"
};
var datasetname = new string[] {
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\09302010_MG1655_phospho_S_08_B_100930113735_msgfdb_fht.txt",
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\10012010_MG1655_phospho_S_09_B_101001105913_msgfdb_fht.txt",
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\10012010_MG1655_phospho_s06_101002094806_msgfdb_fht.txt",
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\10012010_MG1655_phospho_s10_101001105913_msgfdb_fht.txt",
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\10022010_MG1655_phospho_S11_rerun_101005165350_msgfdb_fht.txt",
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\10022010_MG1655_phospho_s12_101002094806_msgfdb_fht.txt",
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\10022010_MG1655_phospho_s7_101002094806_msgfdb_fht.txt"
};
var ascParam = new string[] {
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\HistPhos.xml",
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\HistPhos.xml",
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\HistPhos.xml",
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\HistPhos.xml",
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\HistPhos.xml",
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\HistPhos.xml",
@"C:\Users\aldr699\Documents\2012\Sisi_Work\PhosphohistidineConfirmation\HistPhos.xml"
};
var peptideMassCalculator = GetDefaultPeptideMassCalculator();
for (var i = 0; i < dtaname.Length; i++)
{
var outFile = Path.Combine(
Path.GetDirectoryName(datasetname[i]),
Path.GetFileNameWithoutExtension(datasetname[i]) + "_AScore.txt");
PsmResultsManager datasetMan = new MsgfdbFHT(datasetname[i]);
//DtaManager dtaManager = new DtaManager(dtaname[i]);
var spectraCache = new SpectraManagerCache(peptideMassCalculator);
spectraCache.OpenFile(dtaname[i]);
var paramFile = new ParameterFileManager(ascParam[i]);
var ascoreEngine = new AScore_DLL.AScoreProcessor();
//ascoreEngine.AlgorithmRun(dtaManager, datasetMan, paramFile, outFile);
ascoreEngine.RunAScoreOnSingleFile(spectraCache, datasetMan, paramFile, outFile);
}
}
/// <summary>
/// Configure and run the AScore algorithm, optionally can add protein mapping information
/// </summary>
/// <param name="spectraManager"></param>
/// <param name="psmResultsManager"></param>
/// <param name="ascoreParams"></param>
/// <param name="outputFilePath">Name of the output file</param>
/// <param name="fastaFilePath">Path to FASTA file. If this is empty/null, protein mapping will not occur</param>
/// <param name="outputDescriptions">Whether to include protein description line in output or not.</param>
public void RunAScoreOnSingleFile(
SpectraManagerCache spectraManager,
PsmResultsManager psmResultsManager,
ParameterFileManager ascoreParams,
string outputFilePath,
string fastaFilePath = "",
bool outputDescriptions = false
)
{
var ascoreOptions = new AScoreOptions
{
FastaFilePath = fastaFilePath,
OutputProteinDescriptions = outputDescriptions
};
ascoreOptions.SetAScoreResultsFilePath(outputFilePath);
RunAScoreOnSingleFile(ascoreOptions, spectraManager, psmResultsManager, ascoreParams);
}
/// <summary>
/// Configure and run the AScore algorithm, optionally can add protein mapping information
/// </summary>
/// <param name="ascoreOptions"></param>
/// <param name="spectraManager"></param>
/// <param name="psmResultsManager"></param>
/// <param name="ascoreParams"></param>
public void RunAScoreWithMappingFile(
AScoreOptions ascoreOptions,
SpectraManagerCache spectraManager,
PsmResultsManager psmResultsManager,
ParameterFileManager ascoreParams)
{
var requiredColumns = new List <string>
{
"Job",
"Dataset"
};
OnStatusEvent("Reading Job to Dataset Map File: " + PathUtils.CompactPathString(ascoreOptions.JobToDatasetMapFile, 80));
ReadJobToDatasetMapFile(ascoreOptions, requiredColumns, out var jobToDatasetNameMap);
RunAScoreOnPreparedData(jobToDatasetNameMap, spectraManager, psmResultsManager, ascoreParams, ascoreOptions, false);
ProteinMapperTestRun(ascoreOptions);
}
public void ForJohnJacobs()
{
const string myfht = @"C:\DMS_WorkDir\Step_1_ASCORE\U54_HPp1_LoBMI_NS_11_5Sep08_Draco_08-07-15_xt.txt";
const string mydta = @"C:\DMS_WorkDir\Step_1_ASCORE\dtas\U54_HPp1_LoBMI_NS_11_5Sep08_Draco_08-07-15_dta.txt";
const string mypar = @"C:\DMS_WorkDir\Step_1_ASCORE\DynMWPYOx_EmH20_QmNH3_cid.xml";
var peptideMassCalculator = GetDefaultPeptideMassCalculator();
PsmResultsManager dataman = new XTandemFHT(myfht);
//DtaManager dta = new DtaManager(mydta);
var spectraCache = new SpectraManagerCache(peptideMassCalculator);
spectraCache.OpenFile(mydta);
var par = new ParameterFileManager(mypar);
const string fileOutput = @"C:\DMS_WorkDir\Step_1_ASCORE\U54_HPp1_LoBMI_NS_11_5Sep08_Draco_08-07-15_xt_ascore.txt";
var ascoreEngine = new AScore_DLL.AScoreProcessor();
//ascoreEngine.AlgorithmRun(dta, dataman, par, fileOutput);
ascoreEngine.RunAScoreOnSingleFile(spectraCache, dataman, par, fileOutput);
}
/// <summary>
/// Configure and run the AScore algorithm, optionally can add protein mapping information
/// </summary>
/// <param name="ascoreOptions"></param>
/// <param name="spectraManager"></param>
/// <param name="psmResultsManager"></param>
/// <param name="ascoreParams"></param>
public void RunAScoreOnSingleFile(
AScoreOptions ascoreOptions,
SpectraManagerCache spectraManager,
PsmResultsManager psmResultsManager,
ParameterFileManager ascoreParams)
{
var jobToDatasetNameMap = new Dictionary <string, DatasetFileInfo>
{
{
psmResultsManager.JobNum,
new DatasetFileInfo(spectraManager.SpectrumFilePath, spectraManager.ModSummaryFilePath)
}
};
if (spectraManager == null || !spectraManager.Initialized)
{
throw new Exception(
"spectraManager must be instantiated and initialized before calling RunAScoreOnSingleFile for a single source file");
}
RunAScoreOnPreparedData(jobToDatasetNameMap, spectraManager, psmResultsManager, ascoreParams, ascoreOptions, true);
ProteinMapperTestRun(ascoreOptions);
}
/// <summary>
/// Configure and run the AScore algorithm
/// </summary>
/// <param name="ascoreOptions"></param>
/// <returns></returns>
public int RunAScore(AScoreOptions ascoreOptions)
{
var paramManager = new ParameterFileManager(ascoreOptions.AScoreParamFile);
RegisterEvents(paramManager);
Console.WriteLine();
if (paramManager.DynamicMods.Count > 0 || paramManager.StaticMods.Count > 0)
{
OnStatusEvent("Loaded modifications from: " + ascoreOptions.AScoreParamFile);
foreach (var mod in paramManager.StaticMods)
{
OnStatusEvent(Utilities.GetModDescription("Static, ", mod));
}
foreach (var mod in paramManager.DynamicMods)
{
OnStatusEvent(Utilities.GetModDescription("Dynamic, ", mod));
}
foreach (var mod in paramManager.TerminiMods)
{
OnStatusEvent(Utilities.GetModDescription("Terminus, ", mod));
}
Console.WriteLine();
}
PsmResultsManager psmResultsManager;
switch (ascoreOptions.SearchType)
{
case AScoreOptions.SearchMode.XTandem:
OnStatusEvent("Caching data in " + PathUtils.CompactPathString(ascoreOptions.DbSearchResultsFile, 80));
psmResultsManager = new XTandemFHT(ascoreOptions.DbSearchResultsFile);
break;
case AScoreOptions.SearchMode.Sequest:
OnStatusEvent("Caching data in " + PathUtils.CompactPathString(ascoreOptions.DbSearchResultsFile, 80));
psmResultsManager = new SequestFHT(ascoreOptions.DbSearchResultsFile);
break;
case AScoreOptions.SearchMode.Inspect:
OnStatusEvent("Caching data in " + PathUtils.CompactPathString(ascoreOptions.DbSearchResultsFile, 80));
psmResultsManager = new InspectFHT(ascoreOptions.DbSearchResultsFile);
break;
case AScoreOptions.SearchMode.Msgfdb:
case AScoreOptions.SearchMode.Msgfplus:
OnStatusEvent("Caching data in " + PathUtils.CompactPathString(ascoreOptions.DbSearchResultsFile, 80));
if (ascoreOptions.SearchResultsType == AScoreOptions.DbSearchResultsType.Mzid)
{
if (ascoreOptions.CreateUpdatedDbSearchResultsFile)
{
psmResultsManager = new MsgfMzidFull(ascoreOptions.DbSearchResultsFile);
}
else
{
psmResultsManager = new MsgfMzid(ascoreOptions.DbSearchResultsFile);
}
}
else
{
psmResultsManager = new MsgfdbFHT(ascoreOptions.DbSearchResultsFile);
}
break;
default:
OnErrorEvent(string.Format(
"Incorrect search type: {0} , supported values are {1}",
ascoreOptions.SearchType,
string.Join(", ", Enum.GetNames(typeof(AScoreOptions.SearchMode)))
));
return(-13);
}
var peptideMassCalculator = new PeptideMassCalculator();
var spectraManager = new SpectraManagerCache(peptideMassCalculator);
RegisterEvents(spectraManager);
OnStatusEvent("Output directory: " + ascoreOptions.OutputDirectoryInfo.FullName);
var ascoreEngine = new AScoreAlgorithm();
RegisterEvents(ascoreEngine);
// Initialize the options
FilterOnMSGFScore = ascoreOptions.FilterOnMSGFScore;
// Run the algorithm
if (ascoreOptions.MultiJobMode)
{
RunAScoreWithMappingFile(ascoreOptions, spectraManager, psmResultsManager, paramManager);
}
else
{
spectraManager.OpenFile(ascoreOptions.MassSpecFile, ascoreOptions.ModSummaryFile);
RunAScoreOnSingleFile(ascoreOptions, spectraManager, psmResultsManager, paramManager);
}
OnStatusEvent("AScore Complete");
if (ascoreOptions.CreateUpdatedDbSearchResultsFile)
{
if (ascoreOptions.SearchResultsType == AScoreOptions.DbSearchResultsType.Fht)
{
CreateUpdatedFirstHitsFile(ascoreOptions);
}
else if (psmResultsManager is MsgfMzidFull mzidFull)
{
mzidFull.WriteToMzidFile(ascoreOptions.UpdatedDbSearchResultsFileName);
OnStatusEvent("Results merged; new file: " + PathUtils.CompactPathString(ascoreOptions.UpdatedDbSearchResultsFileName, 80));
}
}
return(0);
}
/// <summary>
/// Runs the all the tools necessary to perform an ascore run
/// </summary>
/// <param name="jobToDatasetNameMap">Keys are job numbers (stored as strings); values are Dataset Names or the path to the _dta.txt file</param>
/// <param name="spectraManager">Manager for reading _dta.txt or .mzML files; must have already been initialized by the calling class</param>
/// <param name="psmResultsManager"></param>
/// <param name="ascoreParams"></param>
/// <param name="ascoreOptions"></param>
/// <param name="spectraFileOpened">Set to true if processing a single dataset, and spectraManager.OpenFile() has already been called</param>
private void RunAScoreOnPreparedData(
IReadOnlyDictionary <string, DatasetFileInfo> jobToDatasetNameMap,
SpectraManagerCache spectraManager,
PsmResultsManager psmResultsManager,
ParameterFileManager ascoreParams,
AScoreOptions ascoreOptions,
bool spectraFileOpened)
{
var totalRows = psmResultsManager.GetRowLength();
var dctPeptidesProcessed = new Dictionary <string, int>();
if (jobToDatasetNameMap == null || jobToDatasetNameMap.Count == 0)
{
const string errorMessage = "Error in AlgorithmRun: jobToDatasetNameMap cannot be null or empty";
OnErrorEvent(errorMessage);
throw new ArgumentException(errorMessage);
}
ISpectraManager spectraFile = null;
string spectraManagerCurrentJob = null; // Force open after first read from fht
var modSummaryManager = new ModSummaryFileManager();
RegisterEvents(modSummaryManager);
var peptideMassCalculator = new PeptideMassCalculator();
if (FilterOnMSGFScore)
{
OnStatusEvent("Filtering using MSGF_SpecProb <= " + ascoreParams.MSGFPreFilter.ToString("0.0E+00"));
}
Console.WriteLine();
var statsByType = new int[4];
var ascoreAlgorithm = new AScoreAlgorithm();
RegisterEvents(ascoreAlgorithm);
while (psmResultsManager.CurrentRowNum < totalRows)
{
// Console.Clear();
if (psmResultsManager.CurrentRowNum % 100 == 0)
{
Console.Write("\rPercent Completion " + Math.Round((double)psmResultsManager.CurrentRowNum / totalRows * 100) + "%");
}
int scanNumber;
int scanCount;
int chargeState;
string peptideSeq;
double msgfScore;
if (FilterOnMSGFScore)
{
psmResultsManager.GetNextRow(out scanNumber, out scanCount, out chargeState, out peptideSeq, out msgfScore, ref ascoreParams);
}
else
{
psmResultsManager.GetNextRow(out scanNumber, out scanCount, out chargeState, out peptideSeq, ref ascoreParams);
msgfScore = 1;
}
switch (ascoreParams.FragmentType)
{
case FragmentType.CID:
statsByType[(int)FragmentType.CID]++;
break;
case FragmentType.ETD:
statsByType[(int)FragmentType.ETD]++;
break;
case FragmentType.HCD:
statsByType[(int)FragmentType.HCD]++;
break;
default:
statsByType[(int)FragmentType.Unspecified]++;
break;
}
if (string.IsNullOrEmpty(spectraManagerCurrentJob) || !string.Equals(spectraManagerCurrentJob, psmResultsManager.JobNum))
{
// New dataset
// Get the correct spectrum file for the match
if (!jobToDatasetNameMap.TryGetValue(psmResultsManager.JobNum, out var datasetInfo))
{
var errorMessage = "Input file refers to job " + psmResultsManager.JobNum +
" but jobToDatasetNameMap does not contain that job; unable to continue";
OnWarningEvent(errorMessage);
if (!psmResultsManager.JobColumnDefined)
{
OnWarningEvent(
"If the input file includes results from multiple jobs, the first column must be job number with Job as the column heading");
}
throw new Exception(errorMessage);
}
var datasetName = GetDatasetName(datasetInfo.SpectrumFilePath);
OnStatusEvent("Dataset name: " + datasetName);
if (!spectraFileOpened)
{
// This method was called from RunAScoreWithMappingFile
// Open the spectrum file for this dataset
spectraFile = spectraManager.GetSpectraManagerForFile(
psmResultsManager.PSMResultsFilePath,
datasetName,
datasetInfo.ModSummaryFilePath);
}
else
{
spectraFile = spectraManager.GetCurrentSpectrumManager();
}
spectraManagerCurrentJob = string.Copy(psmResultsManager.JobNum);
Console.Write("\r");
if (string.IsNullOrWhiteSpace(datasetInfo.ModSummaryFilePath) && !string.IsNullOrWhiteSpace(ascoreOptions.ModSummaryFile))
{
datasetInfo.ModSummaryFilePath = ascoreOptions.ModSummaryFile;
}
if (psmResultsManager is MsgfMzid mzid)
{
mzid.SetModifications(ascoreParams);
}
else if (psmResultsManager is MsgfMzidFull mzidFull)
{
mzidFull.SetModifications(ascoreParams);
}
else
{
if (string.IsNullOrEmpty(datasetInfo.ModSummaryFilePath))
{
modSummaryManager.ReadModSummary(spectraFile.DatasetName, psmResultsManager.PSMResultsFilePath, ascoreParams);
}
else
{
var modSummaryFile = new FileInfo(datasetInfo.ModSummaryFilePath);
modSummaryManager.ReadModSummary(modSummaryFile, ascoreParams);
}
}
Console.WriteLine();
Console.Write("\rPercent Completion " + Math.Round((double)psmResultsManager.CurrentRowNum / totalRows * 100) + "%");
}
// perform work on the match
var peptideParts = peptideSeq.Split('.');
string sequenceWithoutSuffixOrPrefix;
string front;
string back;
if (peptideParts.Length >= 3)
{
front = peptideParts[0];
sequenceWithoutSuffixOrPrefix = peptideParts[1];
back = peptideParts[2];
}
else
{
front = "?";
sequenceWithoutSuffixOrPrefix = string.Copy(peptideSeq);
back = "?";
}
var sequenceClean = GetCleanSequence(sequenceWithoutSuffixOrPrefix, ref ascoreParams);
var skipPSM = FilterOnMSGFScore && msgfScore > ascoreParams.MSGFPreFilter;
var scanChargePeptide = scanNumber + "_" + chargeState + "_" + sequenceWithoutSuffixOrPrefix;
if (dctPeptidesProcessed.ContainsKey(scanChargePeptide))
{
// We have already processed this PSM
skipPSM = true;
}
else
{
dctPeptidesProcessed.Add(scanChargePeptide, 0);
}
if (skipPSM)
{
psmResultsManager.IncrementRow();
continue;
}
//Get experimental spectra
if (spectraFile == null)
{
const string errorMessage = "spectraFile is uninitialized in RunAScoreOnPreparedData; this indicates a programming bug";
OnErrorEvent(errorMessage);
throw new Exception(errorMessage);
}
var expSpec = spectraFile.GetExperimentalSpectra(scanNumber, scanCount, chargeState);
if (expSpec == null)
{
OnWarningEvent("Scan " + scanNumber + " not found in spectra file for peptide " + peptideSeq);
psmResultsManager.IncrementRow();
continue;
}
// Assume monoisotopic for both hi res and low res spectra
MolecularWeights.MassType = MassType.Monoisotopic;
// Compute precursor m/z value
var precursorMZ = peptideMassCalculator.ConvoluteMass(expSpec.PrecursorMass, 1, chargeState);
// Set the m/z range
var mzMax = maxRange;
var mzMin = precursorMZ * lowRangeMultiplier;
if (ascoreParams.FragmentType != FragmentType.CID)
{
mzMax = maxRange;
mzMin = minRange;
}
//Generate all combination mixtures
var modMixture = new Combinatorics.ModMixtureCombo(ascoreParams.DynamicMods, sequenceClean);
var myPositionsList = GetMyPositionList(sequenceClean, modMixture);
//If I have more than 1 modifiable site proceed to calculation
if (myPositionsList.Count > 1)
{
ascoreAlgorithm.ComputeAScore(psmResultsManager, ascoreParams, scanNumber, chargeState,
peptideSeq, front, back, sequenceClean, expSpec,
mzMax, mzMin, myPositionsList);
}
else if (myPositionsList.Count == 1)
{
// Either one or no modifiable sites
var uniqueID = myPositionsList[0].Max();
if (uniqueID == 0)
{
psmResultsManager.WriteToTable(peptideSeq, scanNumber, 0, myPositionsList[0], MOD_INFO_NO_MODIFIED_RESIDUES);
}
else
{
psmResultsManager.WriteToTable(peptideSeq, scanNumber, 0, myPositionsList[0], LookupModInfoByID(uniqueID, ascoreParams.DynamicMods));
}
}
else
{
// No modifiable sites
psmResultsManager.WriteToTable(peptideSeq, scanNumber, 0, new int[0], MOD_INFO_NO_MODIFIED_RESIDUES);
}
psmResultsManager.IncrementRow();
}
Console.WriteLine();
OnStatusEvent(string.Format("Writing {0:N0} rows to {1}", psmResultsManager.ResultsCount, PathUtils.CompactPathString(ascoreOptions.AScoreResultsFilePath, 80)));
psmResultsManager.WriteToFile(ascoreOptions.AScoreResultsFilePath);
Console.WriteLine();
if (statsByType.Sum() == 0)
{
OnWarningEvent("Input file appeared empty");
}
else
{
OnStatusEvent("Stats by fragmentation ion type:");
ReportStatsForFragType(" CID", statsByType, FragmentType.CID);
ReportStatsForFragType(" ETD", statsByType, FragmentType.ETD);
ReportStatsForFragType(" HCD", statsByType, FragmentType.HCD);
}
Console.WriteLine();
}
public void Sisi_Kidneys()
{
const string ascoreP = "HCDPhos.xml";
const string directSeq = @"C:\Users\aldr699\Documents\2012\Sisi_Work\DarthVehDas\Sequest";
const string directMsg = @"C:\Users\aldr699\Documents\2012\Sisi_Work\DarthVehDas\MSGFDB";
var fhtFiles = new List <string> {
"Kidney_ACHN_Das_1_pTyr_HCD_8May12_Lynx_12-02-29_fht.txt",
"Kidney_ACHN_Veh_1_pTyr_HCD_10May12_Lynx_12-02-29_fht.txt",
"Kidney_ACHN_Veh_2_IMAC_HCD_10May12_Lynx_12-02-31_fht.txt",
"Kidney_ACHN_Das_2_IMAC_HCD_11May12_Lynx_12-02-31_fht.txt"
};
var msgfdbFile = new List <string> {
"Kidney_ACHN_Das_1_pTyr_HCD_8May12_Lynx_12-02-29_msgfdb_fht.txt",
"Kidney_ACHN_Veh_1_pTyr_HCD_10May12_Lynx_12-02-29_msgfdb_fht.txt",
"Kidney_ACHN_Veh_2_IMAC_HCD_10May12_Lynx_12-02-31_msgfdb_fht.txt",
"Kidney_ACHN_Das_2_IMAC_HCD_11May12_Lynx_12-02-31_msgfdb_fht.txt"
};
const string direct2 = @"C:\Users\aldr699\Documents\2012\Sisi_Work\DarthVehDas";
var dtaFiles = new List <string> {
"Kidney_ACHN_Das_1_pTyr_HCD_8May12_Lynx_12-02-29_dta.txt",
"Kidney_ACHN_Veh_1_pTyr_HCD_10May12_Lynx_12-02-29_dta.txt",
"Kidney_ACHN_Veh_2_IMAC_HCD_10May12_Lynx_12-02-31_dta.txt",
"Kidney_ACHN_Das_2_IMAC_HCD_11May12_Lynx_12-02-31_dta.txt"
};
var peptideMassCalculator = GetDefaultPeptideMassCalculator();
for (var j = 1; j < 2; j++)
{
for (var i = 0; i < 4; i++)
{
string fht;
string direct;
if (j == 0)
{
direct = directSeq;
fht = fhtFiles[i];
}
else
{
direct = directMsg;
fht = msgfdbFile[i];
}
var tempfht = Path.Combine(direct, fht);
var tempdta = Path.Combine(direct2, dtaFiles[i]);
var ascP = Path.Combine(direct2, ascoreP);
var tempout = Path.Combine(direct, fhtFiles[i].Substring(0, fhtFiles[i].Length - 4) + "_AScore.txt");
PsmResultsManager dsman;
if (j == 0)
{
dsman = new SequestFHT(tempfht);
}
else
{
dsman = new MsgfdbFHT(tempfht);
}
if (dsman == null)
{
Console.WriteLine("Failed to load fht");
}
//DtaManager dtman = new DtaManager(tempdta);
var spectraCache = new SpectraManagerCache(peptideMassCalculator);
spectraCache.OpenFile(tempdta);
var pman = new ParameterFileManager(ascP);
var ascoreEngine = new AScore_DLL.AScoreProcessor();
//ascoreEngine.AlgorithmRun(dtman, dsman, pman, tempout);
ascoreEngine.RunAScoreOnSingleFile(spectraCache, dsman, pman, tempout);
}
}
}
public void OsmaniRedux()
{
const string ascorePETD = "ETDPhos.xml";
const string ascorePCID = "CIDPhos.xml";
const string directETD = @"C:\Users\aldr699\Documents\2012\Osmani\ETD";
const string directCID = @"C:\Users\aldr699\Documents\2012\Osmani\CID";
var fhtFiles = new List <string> {
"SOSM_May_G2_RR1_26Jan12_Hawk_11-11-03p_fhtf.txt",
"SOSM_May_M_RR1_27Jan12_Hawk_11-11-03p_fhtf.txt",
"SOSM_May_G2_RR2_27Jan12_Hawk_11-11-03p_fhtf.txt",
"SOSM_May_M_RR2_29Jan12_Hawk_11-11-03p_fhtf.txt",
"SOSM_May_G_RR1_31Jan12_Hawk_11-11-03p_fhtf.txt",
"SOSM_May_P1_RR1_31Jan12_Hawk_11-11-03p_fhtf.txt"
};
const string direct2 = @"C:\Users\aldr699\Documents\2012\Osmani";
var dtaFiles = new List <string> {
"SOSM_May_G2_RR1_26Jan12_Hawk_11-11-03p_dta.txt",
"SOSM_May_M_RR1_27Jan12_Hawk_11-11-03p_dta.txt",
"SOSM_May_G2_RR2_27Jan12_Hawk_11-11-03p_dta.txt",
"SOSM_May_M_RR2_29Jan12_Hawk_11-11-03p_dta.txt",
"SOSM_May_G_RR1_31Jan12_Hawk_11-11-03p_dta.txt",
"SOSM_May_P1_RR1_31Jan12_Hawk_11-11-03p_dta.txt"
};
var peptideMassCalculator = GetDefaultPeptideMassCalculator();
for (var j = 0; j < 2; j++)
{
for (var i = 0; i < fhtFiles.Count; i++)
{
string direct;
string ascoreP;
if (j == 0)
{
direct = directETD;
ascoreP = ascorePETD;
}
else
{
direct = directCID;
ascoreP = ascorePCID;
}
var tempfht = Path.Combine(direct, fhtFiles[i]);
var tempdta = Path.Combine(direct2, dtaFiles[i]);
var ascP = Path.Combine(direct, ascoreP);
var tempout = Path.Combine(direct, fhtFiles[i].Substring(0, fhtFiles[i].Length - 4) + "_AScore.txt");
PsmResultsManager dsman = new SequestFHT(tempfht);
//DtaManager dtman = new DtaManager(tempdta);
var spectraCache = new SpectraManagerCache(peptideMassCalculator);
spectraCache.OpenFile(tempdta);
var pman = new ParameterFileManager(ascP);
var ascoreEngine = new AScore_DLL.AScoreProcessor();
//ascoreEngine.AlgorithmRun(dtman, dsman, pman, tempout);
ascoreEngine.RunAScoreOnSingleFile(spectraCache, dsman, pman, tempout);
}
}
}
public void QuYi()
{
const string direct = @"C:\Documents and Settings\aldr699\My Documents2011\EColiPhos";
const string ascoreP = "parameterFile.xml";
var fhtFiles = new List <string> {
"775945_10012010_MG1655_phospho_s05_101002094806_msgfdb_fht.txt",
"775946_10012010_MG1655_phospho_s11_101002094806_msgfdb_fht.txt",
"775947_10022010_MG1655_phospho_s7_101002094806_msgfdb_fht.txt",
"775948_10012010_MG1655_phospho_s10_101001105913_msgfdb_fht.txt",
"775949_10012010_MG1655_phospho_s06_101002094806_msgfdb_fht.txt",
"775950_10022010_MG1655_phospho_s12_101002094806_msgfdb_fht.txt",
"775951_10022010_MG1655_phospho_S8FT_101005165350_msgfdb_fht.txt",
"775944_10012010_MG1655_phospho_S_09_B_101001105913_msgfdb_fht.txt",
"775942_09302010_MG1655_phospho_S_08_B_100930113735_msgfdb_fht.txt",
"775952_10022010_MG1655_phospho_S11_rerun_101005165350_msgfdb_fht.txt",
"775955_MG1655_phospho_S-01FT_5Nov10_Hawk_10-11-02p_msgfdb_fht.txt",
"775956_MG1655_phospho_S-02FT_5Nov10_Hawk_10-11-02p_msgfdb_fht.txt",
"775957_MG1655_phospho_S-03FT_8Nov10_Hawk_10-11-02p_msgfdb_fht.txt",
"775958_MG1655_phospho_S-06FT_8Nov10_Hawk_10-11-02p_msgfdb_fht.txt",
"775959_MG1655_phospho_S-09FT_12Nov10_Hawk_10-11-02p_msgfdb_fht.txt",
"775960_MG1655_phospho_S-10FT_12Nov10_Hawk_10-11-02p_msgfdb_fht.txt",
"775961_MG1655_phospho_S-12FT_15Nov10_Hawk_10-11-02p_msgfdb_fht.txt",
"775962_MG1655_phospho_S-01_15Nov10_Hawk_10-11-02p_msgfdb_fht.txt",
"775963_MG1655_phospho_S-02_17Nov10_Hawk_10-11-02p_msgfdb_fht.txt",
"775964_MG1655_phospho_S-03_17Nov10_Hawk_10-11-02p_msgfdb_fht.txt",
"775965_MG1655_phospho_S-04_17Nov10_Hawk_10-11-02p_msgfdb_fht.txt",
"775966_MG1655_phospho_S-07_17Nov10_Hawk_10-11-02p_msgfdb_fht.txt",
"775967_MG1655_phospho_S-12_19Nov10_Hawk_10-11-02p_msgfdb_fht.txt",
"775968_MG1655_phospho_S-10_19Nov10_Hawk_10-11-02p_msgfdb_fht.txt",
"775969_MG1655_phospho_S-09_19Nov10_Hawk_10-11-02p_msgfdb_fht.txt",
"775953_10022010_MG1655_phospho_S05_rerun_101005165350_msgfdb_fht.txt",
"775943_09302010_MG1655_phospho_S_04_B_100930113735_msgfdb_fht.txt",
"775954_10022010_MG1655_phospho_S06_rerun_101005165350_msgfdb_fht.txt"
};
var dtaFiles = new List <string> {
"10012010_MG1655_phospho_s05_101002094806_dta.txt",
"10012010_MG1655_phospho_s11_101002094806_dta.txt",
"10022010_MG1655_phospho_s7_101002094806_dta.txt",
"10012010_MG1655_phospho_s10_101001105913_dta.txt",
"10012010_MG1655_phospho_s06_101002094806_dta.txt",
"10022010_MG1655_phospho_s12_101002094806_dta.txt",
"10022010_MG1655_phospho_S8FT_101005165350_dta.txt",
"10012010_MG1655_phospho_S_09_B_101001105913_dta.txt",
"09302010_MG1655_phospho_S_08_B_100930113735_dta.txt",
"10022010_MG1655_phospho_S11_rerun_101005165350_dta.txt",
"MG1655_phospho_S-01FT_5Nov10_Hawk_10-11-02p_dta.txt",
"MG1655_phospho_S-02FT_5Nov10_Hawk_10-11-02p_dta.txt",
"MG1655_phospho_S-03FT_8Nov10_Hawk_10-11-02p_dta.txt",
"MG1655_phospho_S-06FT_8Nov10_Hawk_10-11-02p_dta.txt",
"MG1655_phospho_S-09FT_12Nov10_Hawk_10-11-02p_dta.txt",
"MG1655_phospho_S-10FT_12Nov10_Hawk_10-11-02p_dta.txt",
"MG1655_phospho_S-12FT_15Nov10_Hawk_10-11-02p_dta.txt",
"MG1655_phospho_S-01_15Nov10_Hawk_10-11-02p_dta.txt",
"MG1655_phospho_S-02_17Nov10_Hawk_10-11-02p_dta.txt",
"MG1655_phospho_S-03_17Nov10_Hawk_10-11-02p_dta.txt",
"MG1655_phospho_S-04_17Nov10_Hawk_10-11-02p_dta.txt",
"MG1655_phospho_S-07_17Nov10_Hawk_10-11-02p_dta.txt",
"MG1655_phospho_S-12_19Nov10_Hawk_10-11-02p_dta.txt",
"MG1655_phospho_S-10_19Nov10_Hawk_10-11-02p_dta.txt",
"MG1655_phospho_S-09_19Nov10_Hawk_10-11-02p_dta.txt",
"10022010_MG1655_phospho_S05_rerun_101005165350_dta.txt",
"09302010_MG1655_phospho_S_04_B_100930113735_dta.txt",
"10022010_MG1655_phospho_S06_rerun_101005165350_dta.txt"
};
var peptideMassCalculator = GetDefaultPeptideMassCalculator();
for (var i = 0; i < fhtFiles.Count; i++)
{
var tempfht = Path.Combine(direct, fhtFiles[i]);
var tempdta = Path.Combine(direct, dtaFiles[i]);
var ascP = Path.Combine(direct, ascoreP);
var tempout = Path.Combine(direct, fhtFiles[i].Substring(0, fhtFiles[i].Length - 4) + "_AScore.txt");
PsmResultsManager dsman = new MsgfdbFHT(tempfht);
//DtaManager dtman = new DtaManager(tempdta);
var spectraCache = new SpectraManagerCache(peptideMassCalculator);
spectraCache.OpenFile(tempdta);
var pman = new ParameterFileManager(ascP);
var ascoreEngine = new AScore_DLL.AScoreProcessor();
//ascoreEngine.AlgorithmRun(dtman, dsman, pman, tempout);
ascoreEngine.RunAScoreOnSingleFile(spectraCache, dsman, pman, tempout);
}
}
