public void TestFeatureAlignment()
{
const string outFilePath = @"\\protoapps\UserData\Jungkap\Lewy\aligned\promex_crosstab_temp.tsv";
//CPTAC_Intact_CR32A_24Aug15_Bane_15-02-06-RZ
var prsmReader = new ProteinSpectrumMatchReader();
var tolerance = new Tolerance(10);
var alignment = new LcMsFeatureAlignment(new AnalysisCompRef.CompRefFeatureComparer(tolerance));
for (var i = 0; i < NdataSet; i++)
{
var rawFile = string.Format(@"{0}\{1}.pbf", PbfPath, GetDataSetNames(i));
var mspFile = string.Format(@"{0}\{1}_IcTda.tsv", MsPfFolder, GetDataSetNames(i));
var mspFile2 = string.Format(@"{0}\{1}_IcTda.tsv", MsPfFolder2, GetDataSetNames(i));
var ms1FtFile = string.Format(@"{0}\{1}.ms1ft", Ms1FtFolder, GetDataSetNames(i));
Console.WriteLine(rawFile);
var run = PbfLcMsRun.GetLcMsRun(rawFile);
var prsmList1 = prsmReader.LoadIdentificationResult(mspFile, ProteinSpectrumMatch.SearchTool.MsPathFinder);
var prsmList2 = prsmReader.LoadIdentificationResult(mspFile2, ProteinSpectrumMatch.SearchTool.MsPathFinder);
prsmList1.AddRange(prsmList2);
var prsmList = MergePrsm(prsmList1);
var features = LcMsFeatureAlignment.LoadProMexResult(i, ms1FtFile, run);
for (var j = 0; j < prsmList.Count; j++)
{
var match = prsmList[j];
match.ProteinId = match.ProteinName;
}
// tag features by PrSMs
for (var j = 0; j < features.Count; j++)
{
//features[j].ProteinSpectrumMatches = new ProteinSpectrumMatchSet(i);
var massTol = tolerance.GetToleranceAsTh(features[j].Mass);
foreach (var match in prsmList)
{
if (features[j].MinScanNum < match.ScanNum && match.ScanNum < features[j].MaxScanNum && Math.Abs(features[j].Mass - match.Mass) < massTol)
{
features[j].ProteinSpectrumMatches.Add(match);
}
}
}
alignment.AddDataSet(i, features, run);
}
alignment.AlignFeatures();
Console.WriteLine("{0} alignments ", alignment.CountAlignedFeatures);
for (var i = 0; i < NdataSet; i++)
{
alignment.FillMissingFeatures(i);
Console.WriteLine("{0} has been processed", GetDataSetNames(i));
}
OutputCrossTabWithId(outFilePath, alignment);
}
public void TestFeatureAlignment()
{
const string outFilePath = @"\\protoapps\UserData\Jungkap\CompRef\aligned\promex_crosstab_temp.tsv";
var runLabels = new[] { "32A", "32B", "32C", "32D", "32E", "32F", "32G", "33A", "33B", "33C", "33D", "33E", "33F", "33G" };
var nDataset = runLabels.Length;
var prsmReader = new ProteinSpectrumMatchReader();
var tolerance = new Tolerance(10);
var alignment = new LcMsFeatureAlignment(new CompRefFeatureComparer(tolerance));
for (var i = 0; i < nDataset; i++)
{
var rawFile = string.Format(@"{0}\CPTAC_Intact_CR{1}_24Aug15_Bane_15-02-06-RZ.pbf", RawFolder, runLabels[i]);
var mspFile = string.Format(@"{0}\CPTAC_Intact_CR{1}_24Aug15_Bane_15-02-06-RZ_IcTda.tsv", MsPfFolder, runLabels[i]);
var ms1FtFile = string.Format(@"{0}\CPTAC_Intact_CR{1}_24Aug15_Bane_15-02-06-RZ.ms1ft", Ms1FtFolder, runLabels[i]);
var run = PbfLcMsRun.GetLcMsRun(rawFile);
var features = LcMsFeatureAlignment.LoadProMexResult(i, ms1FtFile, run);
if (File.Exists(mspFile))
{
var prsmList = prsmReader.LoadIdentificationResult(mspFile, ProteinSpectrumMatch.SearchTool.MsPathFinder);
for (var j = 0; j < prsmList.Count; j++)
{
var match = prsmList[j];
match.ProteinId = match.ProteinName;
}
// tag features by PrSMs
for (var j = 0; j < features.Count; j++)
{
//features[j].ProteinSpectrumMatches = new ProteinSpectrumMatchSet(i);
var massTol = tolerance.GetToleranceAsMz(features[j].Mass);
foreach (var match in prsmList)
{
if (features[j].MinScanNum < match.ScanNum && match.ScanNum < features[j].MaxScanNum && Math.Abs(features[j].Mass - match.Mass) < massTol)
{
features[j].ProteinSpectrumMatches.Add(match);
}
}
}
}
alignment.AddDataSet(i, features, run);
}
alignment.AlignFeatures();
Console.WriteLine("{0} alignments ", alignment.CountAlignedFeatures);
for (var i = 0; i < nDataset; i++)
{
alignment.FillMissingFeatures(i);
Console.WriteLine("{0} has been processed", runLabels[i]);
}
OutputCrossTabWithId(outFilePath, alignment, runLabels);
}
private void AlignFeatures(List <string> datasets, string mspfFolder, string ms1ftFolder, string outFilePath)
{
var nDataset = datasets.Count;
var prsmReader = new ProteinSpectrumMatchReader();
var tolerance = new Tolerance(12);
var alignment = new LcMsFeatureAlignment(new AnalysisCompRef.CompRefFeatureComparer(tolerance));
for (var i = 0; i < nDataset; i++)
{
var rawFile = string.Format(@"{0}\{1}.pbf", PbfPath, datasets[i]);
var mspFile = string.Format(@"{0}\{1}_IcTda.tsv", mspfFolder, datasets[i]);
var ms1FtFile = string.Format(@"{0}\{1}.ms1ft", ms1ftFolder, datasets[i]);
var ms1FtFile2 = string.Format(@"{0}\{1}.seqtag.ms1ft", ms1ftFolder, datasets[i]);
var run = PbfLcMsRun.GetLcMsRun(rawFile);
var features = LcMsFeatureAlignment.LoadProMexResult(i, ms1FtFile, run);
var features2 = LcMsFeatureAlignment.LoadProMexResult(i, ms1FtFile2, run);
features.AddRange(features2);
if (File.Exists(mspFile))
{
var prsmList = prsmReader.LoadIdentificationResult(mspFile, ProteinSpectrumMatch.SearchTool.MsPathFinder);
//var prsmFeatureMatch = new bool[prsmList.Count];
foreach (var match in prsmList)
{
match.ProteinId = match.ProteinName;
}
// tag features by PrSMs
foreach (var feature in features)
{
//features[j].ProteinSpectrumMatches = new ProteinSpectrumMatchSet(i);
var massTol = tolerance.GetToleranceAsMz(feature.Mass);
foreach (var match in prsmList)
{
if (feature.MinScanNum < match.ScanNum && match.ScanNum < feature.MaxScanNum && Math.Abs(feature.Mass - match.Mass) < massTol)
{
feature.ProteinSpectrumMatches.Add(match);
//prsmFeatureMatch[k] = true;
}
}
}
}
alignment.AddDataSet(i, features, run);
}
alignment.AlignFeatures();
Console.WriteLine("{0} alignments ", alignment.CountAlignedFeatures);
for (var i = 0; i < nDataset; i++)
{
alignment.FillMissingFeatures(i);
Console.WriteLine("{0} has been processed", datasets[i]);
}
AnalysisCompRef.OutputCrossTabWithId(outFilePath, alignment, datasets);
}
public void TestFeatureAlignment()
{
const string outFilePath = @"\\protoapps\UserData\Jungkap\Lewy\aligned\promex_crosstab_temp.tsv";
//CPTAC_Intact_CR32A_24Aug15_Bane_15-02-06-RZ
var prsmReader = new ProteinSpectrumMatchReader();
var tolerance = new Tolerance(10);
var alignment = new LcMsFeatureAlignment(new AnalysisCompRef.CompRefFeatureComparer(tolerance));
for (var i = 0; i < NdataSet; i++)
{
var rawFile = string.Format(@"{0}\{1}.pbf", PbfPath, GetDataSetNames(i));
var mspFile = string.Format(@"{0}\{1}_IcTda.tsv", MsPfFolder, GetDataSetNames(i));
var mspFile2 = string.Format(@"{0}\{1}_IcTda.tsv", MsPfFolder2, GetDataSetNames(i));
var ms1FtFile = string.Format(@"{0}\{1}.ms1ft", Ms1FtFolder, GetDataSetNames(i));
Console.WriteLine(rawFile);
var run = PbfLcMsRun.GetLcMsRun(rawFile);
var prsmList1 = prsmReader.LoadIdentificationResult(mspFile, ProteinSpectrumMatch.SearchTool.MsPathFinder);
var prsmList2 = prsmReader.LoadIdentificationResult(mspFile2, ProteinSpectrumMatch.SearchTool.MsPathFinder);
prsmList1.AddRange(prsmList2);
var prsmList = MergePrsm(prsmList1);
var features = LcMsFeatureAlignment.LoadProMexResult(i, ms1FtFile, run);
for (var j = 0; j < prsmList.Count; j++)
{
var match = prsmList[j];
match.ProteinId = match.ProteinName;
}
// tag features by PrSMs
for (var j = 0; j < features.Count; j++)
{
//features[j].ProteinSpectrumMatches = new ProteinSpectrumMatchSet(i);
var massTol = tolerance.GetToleranceAsTh(features[j].Mass);
foreach (var match in prsmList)
{
if (features[j].MinScanNum < match.ScanNum && match.ScanNum < features[j].MaxScanNum && Math.Abs(features[j].Mass - match.Mass) < massTol)
{
features[j].ProteinSpectrumMatches.Add(match);
}
}
}
alignment.AddDataSet(i, features, run);
}
alignment.AlignFeatures();
Console.WriteLine("{0} alignments ", alignment.CountAlignedFeatures);
for (var i = 0; i < NdataSet; i++)
{
alignment.FillMissingFeatures(i);
Console.WriteLine("{0} has been processed", GetDataSetNames(i));
}
OutputCrossTabWithId(outFilePath, alignment);
}
public void TestIMERFeatureAlignment()
{
const string outFilePath = @"D:\MassSpecFiles\IMER\promex_crosstab.tsv";
const string rawFolder = @"D:\MassSpecFiles\IMER";
var runLabels = new string[] { "1", "2", "3", "4", "5", "6" };
var nDataset = runLabels.Length;
//CPTAC_Intact_CR32A_24Aug15_Bane_15-02-06-RZ
var prsmReader = new ProteinSpectrumMatchReader();
var tolerance = new Tolerance(10);
var alignment = new LcMsFeatureAlignment(new CompRefFeatureComparer(tolerance));
for (var i = 0; i < nDataset; i++)
{
var k = runLabels[i].Equals("2") || runLabels[i].Equals("3") ? 14 : 13;
var rawFile = string.Format(@"{0}\Diabetes_iPSC_Beta_{1}_IMER_{2}May14_Alder_14-01-33.pbf", rawFolder, runLabels[i], k);
var mspFile = string.Format(@"{0}\Diabetes_iPSC_Beta_{1}_IMER_{2}May14_Alder_14-01-33_msgfdb_syn.txt", rawFolder, runLabels[i], k);
var ms1FtFile = string.Format(@"{0}\Diabetes_iPSC_Beta_{1}_IMER_{2}May14_Alder_14-01-33.ms1ft", rawFolder, runLabels[i], k);
Console.WriteLine(rawFile);
Console.WriteLine(File.Exists(rawFile));
var run = PbfLcMsRun.GetLcMsRun(rawFile);
var features = LcMsFeatureAlignment.LoadProMexResult(i, ms1FtFile, run, 500, 15000);
if (File.Exists(mspFile))
{
var prsmList = prsmReader.LoadIdentificationResult(mspFile, ProteinSpectrumMatch.SearchTool.MsGfPlus);
for (var j = 0; j < prsmList.Count; j++)
{
var match = prsmList[j];
match.ProteinId = match.ProteinName;
}
// tag features by PrSMs
for (var j = 0; j < features.Count; j++)
{
//features[j].ProteinSpectrumMatches = new ProteinSpectrumMatchSet(i);
var massTol = tolerance.GetToleranceAsTh(features[j].Mass);
foreach (var match in prsmList)
{
if (features[j].MinScanNum < match.ScanNum && match.ScanNum < features[j].MaxScanNum && Math.Abs(features[j].Mass - match.Mass) < massTol)
{
features[j].ProteinSpectrumMatches.Add(match);
}
}
}
}
alignment.AddDataSet(i, features, run);
}
alignment.AlignFeatures();
Console.WriteLine("{0} alignments ", alignment.CountAlignedFeatures);
for (var i = 0; i < nDataset; i++)
{
alignment.FillMissingFeatures(i);
Console.WriteLine("{0} has been processed", runLabels[i]);
}
OutputCrossTabWithId(outFilePath, alignment, runLabels);
}
public void AlignFeatures(List<string> datasets, string mspfFolder, string ms1ftFolder, string outFilePath)
{
var nDataset = datasets.Count;
var prsmReader = new ProteinSpectrumMatchReader();
var tolerance = new Tolerance(12);
var alignment = new LcMsFeatureAlignment(new AnalysisCompRef.CompRefFeatureComparer(tolerance));
for (var i = 0; i < nDataset; i++)
{
var rawFile = string.Format(@"{0}\{1}.pbf", PbfPath, datasets[i]);
var mspFile = string.Format(@"{0}\{1}_IcTda.tsv", mspfFolder, datasets[i]);
var ms1FtFile = string.Format(@"{0}\{1}.ms1ft", ms1ftFolder, datasets[i]);
var ms1FtFile2 = string.Format(@"{0}\{1}.seqtag.ms1ft", ms1ftFolder, datasets[i]);
var run = PbfLcMsRun.GetLcMsRun(rawFile);
var features = LcMsFeatureAlignment.LoadProMexResult(i, ms1FtFile, run);
var features2 = LcMsFeatureAlignment.LoadProMexResult(i, ms1FtFile2, run);
features.AddRange(features2);
if (File.Exists(mspFile))
{
var prsmList = prsmReader.LoadIdentificationResult(mspFile, ProteinSpectrumMatch.SearchTool.MsPathFinder);
//var prsmFeatureMatch = new bool[prsmList.Count];
for (var j = 0; j < prsmList.Count; j++)
{
var match = prsmList[j];
match.ProteinId = match.ProteinName;
}
// tag features by PrSMs
for (var j = 0; j < features.Count; j++)
{
//features[j].ProteinSpectrumMatches = new ProteinSpectrumMatchSet(i);
var massTol = tolerance.GetToleranceAsTh(features[j].Mass);
for(var k = 0; k < prsmList.Count; k++)
{
var match = prsmList[k];
if (features[j].MinScanNum < match.ScanNum && match.ScanNum < features[j].MaxScanNum && Math.Abs(features[j].Mass - match.Mass) < massTol)
{
features[j].ProteinSpectrumMatches.Add(match);
//prsmFeatureMatch[k] = true;
}
}
}
}
alignment.AddDataSet(i, features, run);
}
alignment.AlignFeatures();
Console.WriteLine("{0} alignments ", alignment.CountAlignedFeatures);
for (var i = 0; i < nDataset; i++)
{
alignment.FillMissingFeatures(i);
Console.WriteLine("{0} has been processed", datasets[i]);
}
AnalysisCompRef.OutputCrossTabWithId(outFilePath, alignment, datasets.ToArray());
}
public static void Main(string[] args)
{
// Parse file
var inputFilePath = args[0];
var datasets = DatasetInfo.ParseDatasetInfoFile(inputFilePath);
var fileName = Path.GetFileNameWithoutExtension(inputFilePath);
var directory = Path.GetDirectoryName(inputFilePath);
var outputfilePath = Path.Combine(directory, string.Format("{0}_crosstab.tsv", fileName));
int nDataset = datasets.Count;
var prsmReader = new ProteinSpectrumMatchReader();
var tolerance = new Tolerance(100);
var alignment = new LcMsFeatureAlignment(new CompRefFeatureComparer(tolerance));
int dataId = 0;
foreach (var dataset in datasets)
{
var run = PbfLcMsRun.GetLcMsRun(dataset.RawFilePath, 0, 0);
var features = LcMsFeatureAlignment.LoadProMexResult(dataId++, dataset.Ms1FtFilePath, run);
if (File.Exists(dataset.MsPfIdFilePath))
{
var prsmList = prsmReader.LoadIdentificationResult(dataset.MsPfIdFilePath, ProteinSpectrumMatch.SearchTool.MsPathFinder);
foreach (var match in prsmList)
{
match.ProteinId = match.ProteinName;
}
// tag features by PrSMs
foreach (LcMsFeature feature in features)
{
//features[j].ProteinSpectrumMatches = new ProteinSpectrumMatchSet(i);
var massTol = tolerance.GetToleranceAsMz(feature.Mass);
foreach (var match in prsmList)
{
if (feature.MinScanNum < match.ScanNum && match.ScanNum < feature.MaxScanNum && Math.Abs(feature.Mass - match.Mass) < massTol)
{
feature.ProteinSpectrumMatches.Add(match);
}
}
}
}
alignment.AddDataSet(dataId, features, run);
}
alignment.AlignFeatures();
Console.WriteLine("{0} alignments ", alignment.CountAlignedFeatures);
for (var i = 0; i < nDataset; i++)
{
alignment.FillMissingFeatures(i);
Console.WriteLine("{0} has been processed", datasets[i].Label);
}
OutputCrossTabWithId(outputfilePath, alignment, datasets.Select(ds => ds.Label).ToArray());
}
public static void Main(string[] args)
{
if (args.Length == 0)
{
ShowSyntax();
return;
}
// Parse file
var inputFilePath = args[0];
if (!File.Exists(inputFilePath))
{
ConsoleMsgUtils.ShowError("File not found: " + inputFilePath);
return;
}
var datasets = DatasetInfo.ParseDatasetInfoFile(inputFilePath);
if (datasets.Count == 0)
{
ConsoleMsgUtils.ShowError("No valid data found in the dataset info file");
ShowSyntax();
return;
}
var fileName = Path.GetFileNameWithoutExtension(inputFilePath);
var directory = Path.GetDirectoryName(inputFilePath);
var crosstabFilename = string.Format("{0}_crosstab.tsv", fileName);
string outputfilePath;
if (string.IsNullOrWhiteSpace(directory))
{
outputfilePath = crosstabFilename;
}
else
{
outputfilePath = Path.Combine(directory, crosstabFilename);
}
var nDataset = datasets.Count;
var prsmReader = new ProteinSpectrumMatchReader();
var tolerance = new Tolerance(100);
var alignment = new LcMsFeatureAlignment(new CompRefFeatureComparer(tolerance));
var dataId = 0;
foreach (var dataset in datasets)
{
if (!File.Exists(dataset.RawFilePath))
{
ConsoleMsgUtils.ShowError("Instrument file not found: " + dataset.RawFilePath);
continue;
}
if (!File.Exists(dataset.Ms1FtFilePath))
{
ConsoleMsgUtils.ShowError("ProMex results file not found: " + dataset.Ms1FtFilePath);
continue;
}
Console.WriteLine("Opening " + dataset.RawFilePath);
var run = PbfLcMsRun.GetLcMsRun(dataset.RawFilePath, 0, 0);
Console.WriteLine("Opening " + dataset.Ms1FtFilePath);
var features = LcMsFeatureAlignment.LoadProMexResult(dataId++, dataset.Ms1FtFilePath, run);
if (!string.IsNullOrWhiteSpace(dataset.MsPfIdFilePath) && File.Exists(dataset.MsPfIdFilePath))
{
Console.WriteLine("Opening " + dataset.MsPfIdFilePath);
var prsmList = prsmReader.LoadIdentificationResult(dataset.MsPfIdFilePath, ProteinSpectrumMatch.SearchTool.MsPathFinder);
foreach (var match in prsmList)
{
match.ProteinId = match.ProteinName;
}
// tag features by PrSMs
foreach (var feature in features)
{
//features[j].ProteinSpectrumMatches = new ProteinSpectrumMatchSet(i);
var massTol = tolerance.GetToleranceAsMz(feature.Mass);
foreach (var match in prsmList)
{
if (feature.MinScanNum < match.ScanNum && match.ScanNum < feature.MaxScanNum && Math.Abs(feature.Mass - match.Mass) < massTol)
{
feature.ProteinSpectrumMatches.Add(match);
}
}
}
}
alignment.AddDataSet(dataId, features, run);
}
alignment.AlignFeatures();
Console.WriteLine("{0} alignments ", alignment.CountAlignedFeatures);
var validResults = 0;
for (var datasetIndex = 0; datasetIndex < nDataset; datasetIndex++)
{
if (datasetIndex >= alignment.CountDatasets)
{
ConsoleMsgUtils.ShowWarning(string.Format("Could not align {0}; features not found", datasets[datasetIndex].Label));
continue;
}
alignment.FillMissingFeatures(datasetIndex);
Console.WriteLine("{0} has been processed", datasets[datasetIndex].Label);
validResults++;
}
if (validResults > 0)
{
OutputCrossTabWithId(outputfilePath, alignment, datasets.Select(ds => ds.Label).ToArray());
}
}