public void OnExportAllGlobalResults(string fileLocation)
{
IProgress <int> progress = new Progress <int>(ReportGlobalWorkflowProgress);
LipidGroupSearchResultWriter.OutputResults(LipidGroupSearchResultList, fileLocation, Path.GetFileName(RawFilePath), progress);
progress.Report(0);
}
public void RunTraining(string rawDirectoryPath, string targetFilePath, string decoyFilePath, double hcdError = 30, double cidError = 500)
{
// Read target file
var targetReader = new LipidMapsDbReader <Lipid>();
var targets = targetReader.ReadFile(new FileInfo(targetFilePath));
// Read decoy file
var decoyReader = new LipidMapsDbReader <Lipid>();
var decoys = decoyReader.ReadFile(new FileInfo(decoyFilePath));
var files = Directory.GetFiles(rawDirectoryPath);
foreach (var rawFilePath in files.Where(file => file.EndsWith(".raw")))
{
// create output paths
var rawFileName = Path.GetFileName(rawFilePath);
var datasetPath = Path.GetDirectoryName(rawFilePath);
var datasetName = Path.GetFileNameWithoutExtension(rawFilePath);
var targetResultsPath = Path.Combine(datasetPath, string.Format("{0}_target.tsv", datasetName));
var decoyResultsPath = Path.Combine(datasetPath, string.Format("{0}_decoy.tsv", datasetName));
// Run liquid global workflow
var globalWorkflow = new GlobalWorkflow(rawFilePath);
var targetResults = globalWorkflow.RunGlobalWorkflow(targets, hcdError, cidError);
var decoyResults = globalWorkflow.RunGlobalWorkflow(decoys, hcdError, cidError);
// Output results
LipidGroupSearchResultWriter.OutputResults(targetResults, targetResultsPath, rawFileName);
LipidGroupSearchResultWriter.OutputResults(decoyResults, decoyResultsPath, rawFileName);
// Assure that the source data file is closed
globalWorkflow.LcMsRun.Close();
}
}
public void OnExportGlobalResults(string fileLocation)
{
IProgress <int> progress = new Progress <int>(ReportGlobalWorkflowProgress);
var resultsToExport = LipidGroupSearchResultList.Where(x => x.ShouldExport).ToList();
LipidGroupSearchResultWriter.OutputResults(resultsToExport, fileLocation, Path.GetFileName(RawFilePath), progress);
progress.Report(0);
}
/// <summary>
/// Main functionality for running the LIQUID workflow and outputting the results
/// </summary>
/// <param name="targetsFilePath"></param>
/// <param name="outputFileName"></param>
/// <param name="datasetNamesList"></param>
private void RunWorkflowAndOutput(string targetsFilePath, string outputFileName, List <string> datasetNamesList)
{
var targetsFileInfo = new FileInfo(targetsFilePath);
var lipidReader = new LipidMapsDbReader <Lipid>();
var lipidList = lipidReader.ReadFile(targetsFileInfo);
var headerWritten = false;
foreach (var datasetName in datasetNamesList)
{
var rawFileName = datasetName + ".raw";
var rawFilePath = Path.Combine(@"D:\Data\Liquid\Original", rawFileName);
Console.WriteLine(DateTime.Now + ": Processing " + datasetName);
if (File.Exists(rawFilePath))
{
Console.WriteLine(DateTime.Now + ": Dataset already exists");
}
else
{
Console.WriteLine(DateTime.Now + ": Dataset does not exist locally, so we will go get it");
// Lookup in DMS via Mage
var dmsFolder = DmsDatasetFinder.FindLocationOfDataset(datasetName);
var dmsDirectoryInfo = new DirectoryInfo(dmsFolder);
var fullPathToDmsFile = Path.Combine(dmsDirectoryInfo.FullName, rawFileName);
// Copy Locally
// TODO: Handle files that are on MyEMSL
Console.WriteLine(DateTime.Now + ": Copying dataset from " + dmsDirectoryInfo.FullName);
File.Copy(fullPathToDmsFile, rawFilePath);
Console.WriteLine(DateTime.Now + ": Copy complete");
}
// Setup workflow
var globalWorkflow = new GlobalWorkflow(rawFilePath);
// Run workflow
var lipidGroupSearchResults = globalWorkflow.RunGlobalWorkflow(lipidList, 30, 500);
if (!headerWritten)
{
LipidGroupSearchResultWriter.OutputResults(lipidGroupSearchResults, outputFileName, rawFileName, null, true, true);
headerWritten = true;
}
else
{
LipidGroupSearchResultWriter.OutputResults(lipidGroupSearchResults, outputFileName, rawFileName, null, true, false);
}
// Assure that the source data file is closed
globalWorkflow.LcMsRun.Close();
}
}
public void RunTrainingOnFileList(
string fileListPath,
string posTargetFilePath,
string posDecoyFilePath,
string negTargetFilePath,
string negDecoyFilePath,
double hcdError = 30,
double cidError = 500)
{
// Read positive target file
var posTargetReader = new LipidMapsDbReader <Lipid>();
var posTargets = posTargetReader.ReadFile(new FileInfo(posTargetFilePath));
// Read positive decoy file
var posDecoyReader = new LipidMapsDbReader <Lipid>();
var posDecoys = posDecoyReader.ReadFile(new FileInfo(posDecoyFilePath));
// Read positive target file
var negTargetReader = new LipidMapsDbReader <Lipid>();
var negTargets = negTargetReader.ReadFile(new FileInfo(negTargetFilePath));
// Read positive decoy file
var negDecoyReader = new LipidMapsDbReader <Lipid>();
var negDecoys = negDecoyReader.ReadFile(new FileInfo(negDecoyFilePath));
var outputDirectory = Path.GetDirectoryName(fileListPath);
var errorFile = Path.Combine(outputDirectory, "failedDatasets.txt");
foreach (var datasetName in File.ReadLines(fileListPath))
{
if (datasetName.StartsWith("//"))
{
continue;
}
try
{
// create output paths
var rawFilePath = GetRawFilePath(outputDirectory, datasetName);
var rawFileName = Path.GetFileName(rawFilePath);
var targetResultsPath = Path.Combine(outputDirectory, string.Format("{0}_target.tsv", datasetName));
var decoyResultsPath = Path.Combine(outputDirectory, string.Format("{0}_decoy.tsv", datasetName));
IEnumerable <Lipid> targets;
IEnumerable <Lipid> decoys;
// Select targets and decoys
var lowerCaseName = datasetName.ToLower();
if (lowerCaseName.Contains("pos"))
{
targets = posTargets;
decoys = posDecoys;
}
else
{
targets = negTargets;
decoys = negDecoys;
}
// Run liquid global workflow
var globalWorkflow = new GlobalWorkflow(rawFilePath);
var targetResults = GetBestResultPerSpectrum(globalWorkflow.RunGlobalWorkflow(targets, hcdError, cidError));
var decoyResults = GetBestResultPerSpectrum(globalWorkflow.RunGlobalWorkflow(decoys, hcdError, cidError));
// Output results
LipidGroupSearchResultWriter.OutputResults(targetResults, targetResultsPath, rawFileName);
LipidGroupSearchResultWriter.OutputResults(decoyResults, decoyResultsPath, rawFileName);
// Assure that the source data file is closed
globalWorkflow.LcMsRun.Close();
}
catch (Exception)
{
Console.WriteLine("ERROR: Could not process dataset {0}.", datasetName);
using (var streamWriter = new StreamWriter(errorFile, true))
{
streamWriter.WriteLine(datasetName);
}
}
}
}
