/// <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();
}