public void TestTagMatching()
{
var methodName = MethodBase.GetCurrentMethod().Name;
TestUtils.ShowStarting(methodName);
// Parse sequence tags
const string dataSet = @"H:\Research\QCShew_TopDown\Production\QC_Shew_Intact_26Sep14_Bane_C2Column3";
const int minTagLength = 8;
var tagFileName = MassSpecDataReaderFactory.ChangeExtension(dataSet, ".seqtag");
if (!File.Exists(tagFileName))
{
Assert.Ignore(@"Skipping test {0} since file not found: {1}", methodName, tagFileName);
}
var tagParser = new SequenceTagParser(tagFileName, minTagLength);
// Parse raw file
var rawFileName = MassSpecDataReaderFactory.ChangeExtension(dataSet, ".raw");
if (!File.Exists(rawFileName))
{
Assert.Ignore(@"Skipping test {0} since file not found: {1}", methodName, rawFileName);
}
var run = PbfLcMsRun.GetLcMsRun(rawFileName);
// Parse ID file
const string resultFilePath = @"H:\Research\QCShew_TopDown\Production\M1_V092\QC_Shew_Intact_26Sep14_Bane_C2Column3_IcTda.tsv";
if (!File.Exists(resultFilePath))
{
Assert.Ignore(@"Skipping test {0} since file not found: {1}", methodName, resultFilePath);
}
var resultParser = new MsPathFinderParser(resultFilePath);
const double qValueThreshold = 0.01;
var idList = resultParser.GetIdWithQValuesNoLargerThan(qValueThreshold);
var idFlag = new bool[run.MaxLcScan + 1];
foreach (var id in idList) idFlag[id.Scan] = true;
const string fastaFilePath = @"H:\Research\QCShew_TopDown\Production\ID_002216_235ACCEA.fasta";
if (!File.Exists(fastaFilePath))
{
Assert.Ignore(@"Skipping test {0} since file not found: {1}", methodName, fastaFilePath);
}
// const string fastaFilePath = @"H:\Research\QCShew_TopDown\Production\ID_002216_235ACCEA.icsfldecoy.fasta";
var fastaDb = new FastaDatabase(fastaFilePath);
var searchableDb = new SearchableDatabase(fastaDb);
var numMs2Spectra = 0;
var numSpectraWithTag = 0;
var numSpectraWithMatchingTag = 0;
var numSpectraWithMatchedTagNoId = 0;
foreach (var ms2ScanNum in run.GetScanNumbers(2))
{
++numMs2Spectra;
var tags = tagParser.GetSequenceTags(ms2ScanNum);
if (tags != null)
{
++numSpectraWithTag;
foreach (var tag in tags)
{
if (searchableDb.Search(tag.Sequence) >= 0)
{
//Console.WriteLine(tag.Sequence);
++numSpectraWithMatchingTag;
if (!idFlag[ms2ScanNum]) ++numSpectraWithMatchedTagNoId;
break;
}
}
}
}
Console.WriteLine("Tag length: {0}", minTagLength);
Console.WriteLine("NumMs2Spectra: {0}", numMs2Spectra);
Console.WriteLine("NumMs2SpectraWithTags: {0} ({1})", numSpectraWithTag, numSpectraWithTag/(float)numMs2Spectra);
Console.WriteLine("NumMs2SpectraWithMatchingTags: {0} ({1})", numSpectraWithMatchingTag, numSpectraWithMatchingTag/(float)numMs2Spectra);
Console.WriteLine("NumMs2SpectraWithMatchingTagsWithNoId: {0} ({1})", numSpectraWithMatchedTagNoId, numSpectraWithMatchedTagNoId / (float)numMs2Spectra);
}
public void TestFeatureIdMatching()
{
var methodName = MethodBase.GetCurrentMethod().Name;
TestUtils.ShowStarting(methodName);
const string resultFilePath = @"H:\Research\QCShew_TopDown\Production\M1_V092\QC_Shew_Intact_26Sep14_Bane_C2Column3_IcTda.tsv";
if (!File.Exists(resultFilePath))
{
Assert.Ignore(@"Skipping test {0} since file not found: {1}", methodName, resultFilePath);
}
var resultParser = new MsPathFinderParser(resultFilePath);
const double qValueThreshold = 0.01;
const double tolerancePpm = 13;
const string dataSet = @"H:\Research\QCShew_TopDown\Production\QC_Shew_Intact_26Sep14_Bane_C2Column3";
var rawFileName = MassSpecDataReaderFactory.ChangeExtension(dataSet, ".raw");
if (!File.Exists(rawFileName))
{
Assert.Ignore(@"Skipping test {0} since file not found: {1}", methodName, rawFileName);
}
var run = PbfLcMsRun.GetLcMsRun(rawFileName);
var idList =
resultParser.GetIdList().TakeWhile(id => id.QValue <= qValueThreshold).OrderBy(id => id.Mass).ToList();
var idMassList = idList.Select(id => id.Mass).ToList();
var idFlag = new bool[idList.Count];
// Parse sequence tags
var tagFileName = MassSpecDataReaderFactory.ChangeExtension(dataSet, ".seqtag");
const int minTagLength = 6;
const int numProtMatches = 4;
// const string fastaFilePath = @"H:\Research\QCShew_TopDown\Production\ID_002216_235ACCEA.fasta";
const string fastaFilePath = @"H:\Research\QCShew_TopDown\Production\ID_002216_235ACCEA.icsfldecoy.fasta";
if (!File.Exists(tagFileName))
{
Assert.Ignore(@"Skipping test {0} since file not found: {1}", methodName, tagFileName);
}
if (!File.Exists(fastaFilePath))
{
Assert.Ignore(@"Skipping test {0} since file not found: {1}", methodName, fastaFilePath);
}
var fastaDb = new FastaDatabase(fastaFilePath);
var searchableDb = new SearchableDatabase(fastaDb);
var tagParser = new SequenceTagParser(tagFileName, minTagLength);
var featureFileName = MassSpecDataReaderFactory.ChangeExtension(dataSet, ".ms1ft");
var featureParser = new TsvFileParser(featureFileName);
var minScan = featureParser.GetData("MinScan").Select(s => Convert.ToInt32(s)).ToArray();
var maxScan = featureParser.GetData("MaxScan").Select(s => Convert.ToInt32(s)).ToArray();
var minCharge = featureParser.GetData("MinCharge").Select(s => Convert.ToInt32(s)).ToArray();
var maxCharge = featureParser.GetData("MaxCharge").Select(s => Convert.ToInt32(s)).ToArray();
var monoMass = featureParser.GetData("MonoMass").Select(Convert.ToDouble).ToArray();
var numFeaturesWithId = 0;
var numFeaturesWithMs2 = 0;
var numFeaturesWithTags = 0;
var numFeaturesWithMatchingTags = 0;
var numFeaturesWithTwoOrMoreMatchingTags = 0;
var numFeaturesWithNoIdAndMatchingTags = 0;
for (var i = 0; i < featureParser.NumData; i++)
{
var mass = monoMass[i];
// Find Id
var tolDa = new Tolerance(tolerancePpm).GetToleranceAsDa(mass, 1);
var minMass = mass - tolDa;
var maxMass = mass + tolDa;
var index = idMassList.BinarySearch(mass);
if (index < 0) index = ~index;
var matchedId = new List<MsPathFinderId>();
// go down
var curIndex = index - 1;
while (curIndex >= 0)
{
var curId = idList[curIndex];
if (curId.Mass < minMass) break;
if (curId.Scan > minScan[i] && curId.Scan < maxScan[i]
&& curId.Charge >= minCharge[i] && curId.Charge <= maxCharge[i])
{
matchedId.Add(curId);
idFlag[curIndex] = true;
}
--curIndex;
}
// go up
curIndex = index;
while (curIndex < idList.Count)
{
var curId = idList[curIndex];
if (curId.Mass > maxMass) break;
if (curId.Scan >= minScan[i] && curId.Scan <= maxScan[i]
&& curId.Charge >= minCharge[i] && curId.Charge <= maxCharge[i])
{
matchedId.Add(curId);
idFlag[curIndex] = true;
}
++curIndex;
}
var hasId = false;
if (matchedId.Any())
{
++numFeaturesWithId;
hasId = true;
}
// Find MS2 scans
// var numMs2Scans = 0;
var tags = new List<SequenceTag>();
var hasMs2 = false;
for (var scanNum = minScan[i]; scanNum <= maxScan[i]; scanNum++)
{
var isolationWindow = run.GetIsolationWindow(scanNum);
if (isolationWindow == null) continue;
var isolationWindowTargetMz = isolationWindow.IsolationWindowTargetMz;
var charge = (int)Math.Round(mass / isolationWindowTargetMz);
if (charge < minCharge[i] || charge > maxCharge[i]) continue;
var mz = Ion.GetIsotopeMz(mass, charge,
Averagine.GetIsotopomerEnvelope(mass).MostAbundantIsotopeIndex);
if (isolationWindow.Contains(mz))
{
// ++numMs2Scans;
tags.AddRange(tagParser.GetSequenceTags(scanNum));
hasMs2 = true;
}
}
if (hasMs2) ++numFeaturesWithMs2;
if (tags.Any()) ++numFeaturesWithTags;
var protHist = new Dictionary<string, int>();
var hasMatchedTag = false;
foreach (var tag in tags)
{
var matchedProteins = searchableDb.FindAllMatchedSequenceIndices(tag.Sequence).Select(idx => fastaDb.GetProteinName(idx)).ToArray();
if (matchedProteins.Any())
{
hasMatchedTag = true;
foreach (var protein in matchedProteins)
{
int num;
if (protHist.TryGetValue(protein, out num)) protHist[protein] = num + 1;
else protHist[protein] = 1;
}
}
}
if (hasMatchedTag)
{
++numFeaturesWithMatchingTags;
if (!hasId) ++numFeaturesWithNoIdAndMatchingTags;
}
if (protHist.Any())
{
var maxOcc = protHist.Values.Max();
if (maxOcc >= numProtMatches) ++numFeaturesWithTwoOrMoreMatchingTags;
}
}
Console.WriteLine("NumFeatures: {0}", featureParser.NumData);
Console.WriteLine("NumId: {0}", idList.Count);
Console.WriteLine("NumFeaturesWithId: {0} ({1})", numFeaturesWithId, numFeaturesWithId / (float)featureParser.NumData);
Console.WriteLine("NumFeaturesWithMs2: {0} ({1})", numFeaturesWithMs2, numFeaturesWithMs2 / (float)featureParser.NumData);
Console.WriteLine("NumFeaturesWithTag: {0} ({1})", numFeaturesWithTags, numFeaturesWithTags / (float)featureParser.NumData);
Console.WriteLine("NumFeaturesWithMatchedTag: {0} ({1})", numFeaturesWithMatchingTags, numFeaturesWithMatchingTags / (float)featureParser.NumData);
Console.WriteLine("NumFeaturesWithMoreThanOneMatchedTag: {0} ({1})", numFeaturesWithTwoOrMoreMatchingTags, numFeaturesWithTwoOrMoreMatchingTags / (float)featureParser.NumData);
Console.WriteLine("NumFeaturesWithNoIdAndMatchedTag: {0} ({1})", numFeaturesWithNoIdAndMatchingTags, numFeaturesWithNoIdAndMatchingTags / (float)featureParser.NumData);
for (var i = 0; i < idFlag.Length; i++)
{
if (!idFlag[i])
{
Console.WriteLine(idList[i].Scan);
}
}
// Console.WriteLine(string.Join(",", filter.GetMatchingMs2ScanNums(8115.973001)));
//
// Console.WriteLine(featureFileName);
}