public string ProcessFile(string rawFile, string resultFile, string methodName)
{
if (!File.Exists(rawFile))
{
Console.WriteLine(@"Warning: Skipping test {0} since file not found: {1}", methodName, rawFile);
return("\n");
}
if (!File.Exists(resultFile))
{
Console.WriteLine(@"Warning: Skipping test {0} since file not found: {1}", methodName, resultFile);
return("\n");
}
var tsvParser = new TsvFileParser(resultFile);
var headerList = tsvParser.GetHeaders();
var tsvData = tsvParser.GetAllData();
var ms2ScanNumbers = tsvData["Scan"];
var run = PbfLcMsRun.GetLcMsRun(rawFile, 0, 0);
var resultLine = "";
for (int i = 0; i < ms2ScanNumbers.Count; i++)
{
var scanNum = Int32.Parse(ms2ScanNumbers[i]);
var spectrum = run.GetSpectrum(scanNum) as ProductSpectrum;
int tsvIndex = ms2ScanNumbers.FindIndex(x => Int32.Parse(x) == scanNum);
var qValue = Double.Parse(tsvData["QValue"].ElementAt(tsvIndex));
if (qValue > 0.01)
{
continue;
}
var seqStr = tsvData["Sequence"].ElementAt(tsvIndex).Trim();
var seqMod = tsvData["Modifications"].ElementAt(tsvIndex).Trim();
var matchedFrags = tsvData["#MatchedFragments"].ElementAt(tsvIndex).Trim();
var aaSet = new AminoAcidSet();
var sequence = Sequence.CreateSequence(seqStr, seqMod, aaSet);
var tol = new Tolerance(10);
var sequenceFinder = new SequenceTagIndexFinder(tol, 1, 10);
var results = sequenceFinder.GetLongestSequence(spectrum, sequence);
resultLine += String.Format("{0},{1},{2},{3},{4},{5},{6},{7},{8},\n", scanNum, matchedFrags, seqStr, results.Item1, results.Item2, results.Item3, results.Item4, results.Item5, results.Item6);
}
return(resultLine);
}
public void TestScoring()
{
var methodName = MethodBase.GetCurrentMethod().Name;
TestUtils.ShowStarting(methodName);
var rawFile = @"\\protoapps\UserData\Jungkap\Joshua\testData\QC_Shew_Intact_26Sep14_Bane_C2Column3.pbf";
var resultFile = @"\\protoapps\UserData\Jungkap\Joshua\IdResult\QC_Shew_Intact_26Sep14_Bane_C2Column3_IcTda.tsv";
if (!File.Exists(rawFile))
{
Console.WriteLine(@"Warning: Skipping test {0} since file not found: {1}", methodName, rawFile);
return;
}
if (!File.Exists(resultFile))
{
Console.WriteLine(@"Warning: Skipping test {0} since file not found: {1}", methodName, resultFile);
return;
}
var tsvParser = new TsvFileParser(resultFile);
var tsvData = tsvParser.GetAllData();
var ms2ScanNumbers = tsvData["Scan"];
var run = PbfLcMsRun.GetLcMsRun(rawFile, 0, 0);
for (int i = 0; i < 1; i++)
{
var scanNum = Int32.Parse(ms2ScanNumbers[i]);
var spectrum = run.GetSpectrum(scanNum) as ProductSpectrum;
int tsvIndex = ms2ScanNumbers.FindIndex(x => Int32.Parse(x) == scanNum);
var seqStr = tsvData["Sequence"].ElementAt(tsvIndex).Trim();
var seqMod = tsvData["Modifications"].ElementAt(tsvIndex).Trim();
var aaSet = new AminoAcidSet();
var sequence = Sequence.CreateSequence(seqStr, seqMod, aaSet);
Console.WriteLine(sequence.Count);
var score = GetScoreTest(sequence, spectrum);
Console.WriteLine(scanNum + ":" + score);
}
}
public void TestSequenceTag()
{
var methodName = MethodBase.GetCurrentMethod().Name;
Utils.ShowStarting(methodName);
//const string TestRawFile = @"D:\\Vlad_TopDown\\raw\\yufeng_column_test2.raw";
//const string TestResultFile = @"D:\\Vlad_TopDown\\results\\yufeng_column_test2_IcTda.tsv";
const string TestRawFile = @"D:\MassSpecFiles\training\raw\QC_Shew_Intact_26Sep14_Bane_C2Column3.pbf";
const string TestResultFile = @"D:\MassSpecFiles\training\IdResult\QC_Shew_Intact_26Sep14_Bane_C2Column3_IcTda.tsv";
//const string TestRawFile = @"D:\MassSpecFiles\Lewy\Lewy_intact_01.raw";
//const string TestResultFile = @"D:\MassSpecFiles\Lewy\Lewy_intact_01_IcTda.tsv";
if (!File.Exists(TestRawFile))
{
Assert.Ignore(@"Skipping test {0} since file not found: {1}", methodName, TestRawFile);
}
if (!File.Exists(TestResultFile))
{
Assert.Ignore(@"Skipping test {0} since file not found: {1}", methodName, TestResultFile);
}
// Configure amino acid set
var aminoAcidList = new List <AminoAcid>();
foreach (var aa in AminoAcid.StandardAminoAcidArr)
{
aminoAcidList.Add(aa);
aminoAcidList.Add(new ModifiedAminoAcid(aa, Modification.Acetylation));
aminoAcidList.Add(new ModifiedAminoAcid(aa, Modification.Oxidation));
}
//const int MaxTags = 100000;
var tsvParser = new TsvFileParser(TestResultFile);
var headerList = tsvParser.GetHeaders();
var tsvData = tsvParser.GetAllData();
var ms2ScanNumbers = tsvData["Scan"];
var run = PbfLcMsRun.GetLcMsRun(TestRawFile);
var nSpec = 0;
var nHitSpec = 0;
for (var i = 0; i < ms2ScanNumbers.Count; i++)
//foreach(var scanNum in targetScans)
{
var scanNum = Int32.Parse(ms2ScanNumbers[i]);
//if (scanNum != 4672) continue;
var spectrum = run.GetSpectrum(scanNum) as ProductSpectrum;
int tsvIndex = ms2ScanNumbers.FindIndex(x => Int32.Parse(x) == scanNum);
var qValue = double.Parse(tsvData["QValue"].ElementAt(tsvIndex));
if (qValue > 0.01)
{
break;
}
var seqStr = tsvData["Sequence"].ElementAt(tsvIndex).Trim();
var modStr = tsvData["Modifications"].ElementAt(tsvIndex).Trim();
var tolerance = new Tolerance(5);
var tagFinder = new SequenceTagFinder(spectrum, tolerance, 5, 8, aminoAcidList.ToArray());
var nTags = 0;
var nHit = 0;
var seqOjb = Sequence.CreateSequence(seqStr, modStr, new AminoAcidSet());
var compWithoutH2O = seqOjb.Composition - Composition.H2O;
//Console.WriteLine(compWithoutH2O.Mass);
foreach (var seqTagStr in tagFinder.GetAllSequenceTagString())
{
if (seqStr.Contains(seqTagStr.Sequence)) //|| seqStr.Contains(Reverse(tagStr)))
{
//var idx = seqStr.IndexOf(seqTagStr.Sequence);
//seqStr.Substring(0, idx)
/*var comp2 = seqOjb.GetComposition(0, idx);
*
* Console.Write(comp2.Mass);
* Console.Write("\t");
*
* Console.Write(seqTagStr.FlankingMass);
* Console.Write("\t");
* Console.Write(seqTagStr.Sequence);
* Console.Write("\t");
* Console.Write(seqTagStr.IsPrefix);
* Console.WriteLine("");
*/
if (seqStr.Contains(seqTagStr.Sequence))
{
nHit++;
}
}
nTags++;
}
nSpec++;
if (nHit > 0)
{
nHitSpec++;
}
Console.WriteLine(@"[{0}]seqLen = {1}: {2}/{3}", scanNum, seqStr.Length, nHit, nTags);
}
//var existingTags = tagFinder.ExtractExistingSequneceTags(sequence);
Console.Write("{0}/{1}", nHitSpec, nSpec);
}
