private void WriteResultsToFile(DatabaseSequenceSpectrumMatch[] matches, string outputFilePath, FastaDatabase database)
{
using (var writer = new StreamWriter(outputFilePath))
{
writer.WriteLine("Scan\tPre\tSequence\tPost\tModifications\tComposition\tProteinName\tProteinDesc" +
"\tProteinLength\tStart\tEnd\tCharge\tMostAbundantIsotopeMz\tMass\t#MatchedFragments\tProbability\tSpecEValue\tEValue");
foreach (var scanNum in _ms2ScanNums)
{
var match = matches[scanNum];
if (match == null)
{
continue;
}
var sequence = match.Sequence;
var offset = match.Offset;
var start = database.GetOneBasedPositionInProtein(offset) + 1 + match.NumNTermCleavages;
var end = start + sequence.Length - 1;
var proteinName = database.GetProteinName(match.Offset);
var protLength = database.GetProteinLength(proteinName);
var ion = match.Ion;
var proteinDescription = database.GetProteinDescription(match.Offset);
var probability = CompositeScorer.GetProbability(match.Score);
// Note for DblToString(value, 9, true), by having "9" and "true",
// values between 100 and 999 Da will have 7 digits after the decimal place, and
// values between 1000 and 9999 will have 6 digits after the decimal place
writer.WriteLine("{0}\t{1}\t{2}\t{3}\t{4}\t{5}\t{6}\t{7}\t{8}\t{9}\t{10}\t{11}\t{12}\t{13}\t{14}\t{15}\t{16}\t{17}",
scanNum,
match.Pre, // Pre
sequence, // Sequence
match.Post, // Post
match.ModificationText, // Modifications
ion.Composition, // Composition
proteinName, // ProteinName
proteinDescription, // ProteinDescription
protLength, // ProteinLength
start, // Start position in protein
end, // End position in protein
ion.Charge, // precursorCharge
StringUtilities.DblToString(ion.GetMostAbundantIsotopeMz(), 9, true), // MostAbundantIsotopeMz
StringUtilities.DblToString(ion.Composition.Mass, 9, true), // Mass
match.NumMatchedFragments, // (Number of matched fragments)
StringUtilities.DblToString(probability, 4), // Probability
StringUtilities.DblToString(ExcelMinValue(match.SpecEvalue), 6, true, 0.001), // EValue; will be displayed using scientific notation if the value is less than 0.001
StringUtilities.DblToString(ExcelMinValue(match.SpecEvalue * database.GetNumEntries()), 6, true, 0.001) // SpecEValue; will be displayed using scientific notation if the value is less than 0.001
);
}
}
}
private void GetMatchStatistics(ProductSpectrum ms2Spec, Sequence sequence, int parentIonCharge, StreamWriter writer)
{
if (ms2Spec == null)
{
return;
}
if (sequence == null)
{
return;
}
var BaseIonTypesCID = new[] { BaseIonType.B, BaseIonType.Y };
var BaseIonTypesETD = new[] { BaseIonType.C, BaseIonType.Z };
var tolerance = new Tolerance(12);
var MinProductCharge = 1;
var MaxProductCharge = Math.Min(parentIonCharge + 2, 20);
var baseIonTypes = ms2Spec.ActivationMethod != ActivationMethod.ETD ? BaseIonTypesCID : BaseIonTypesETD;
var refIntensity = CompositeScorer.GetRefIntensity(ms2Spec.Peaks);
var activationMethodFlag = ms2Spec.ActivationMethod == ActivationMethod.ETD ? 2 : 1;
var cleavages = sequence.GetInternalCleavages();
var nComplementaryFrags = 0;
var prefixStat = new FragmentStat();
var suffixStat = new FragmentStat();
var minMz = ms2Spec.Peaks.First().Mz;
var maxMz = ms2Spec.Peaks.Last().Mz;
var cleavageIndex = 0;
var preFixIonCheck = new bool[sequence.Count + 1];
var sufFixIonCheck = new bool[sequence.Count + 1];
foreach (var c in cleavages)
{
var prefixHit = false;
var suffixHit = false;
foreach (var baseIonType in baseIonTypes)
{
var stat = baseIonType.IsPrefix ? prefixStat : suffixStat;
var fragmentComposition = baseIonType.IsPrefix
? c.PrefixComposition + baseIonType.OffsetComposition
: c.SuffixComposition + baseIonType.OffsetComposition;
if (fragmentComposition.Mass < ms2Spec.Peaks[0].Mz)
{
continue;
}
var curFragMass = fragmentComposition.Mass;
/*var curObsIonCharge = 0;
* var curObsIonDist = 1.0d;
* var curObsIonCorr = 0d;
* var curObsIonIntensity = 0d;
* var curObsIonMassError = 0d;*/
var mostAbundantIsotopeIndex = fragmentComposition.GetMostAbundantIsotopeZeroBasedIndex();
var fragmentIonMostAbuMass = fragmentComposition.Mass + Constants.C13MinusC12 * mostAbundantIsotopeIndex;
var maxCharge = (int)Math.Floor(fragmentIonMostAbuMass / (minMz - Constants.Proton));
var minCharge = (int)Math.Ceiling(fragmentIonMostAbuMass / (maxMz - Constants.Proton));
if (maxCharge < 1 || maxCharge > MaxProductCharge)
{
maxCharge = MaxProductCharge;
}
if (minCharge < 1 || minCharge < MinProductCharge)
{
minCharge = MinProductCharge;
}
//var ionMatch = false;
for (var charge = minCharge; charge <= maxCharge; charge++)
{
var ion = new Ion(fragmentComposition, charge);
var isotopePeaks = ms2Spec.GetAllIsotopePeaks(ion, tolerance, 0.1);
if (isotopePeaks == null)
{
continue;
}
var distCorr = CompositeScorer.GetDistCorr(ion, isotopePeaks);
if (distCorr.Item2 < 0.7 && distCorr.Item1 > 0.03)
{
continue;
}
var mostAbuPeak = isotopePeaks[mostAbundantIsotopeIndex];
var intScore = mostAbuPeak.Intensity / refIntensity;
/*
* if (ionMatch == false || curObsIonIntensity < intScore)
* {
* curObsIonCharge = charge;
* curObsIonCorr = distCorr.Item2;
* curObsIonDist = distCorr.Item1;
* curObsIonIntensity = intScore;
*
* var mostAbuPeakMz = Ion.GetIsotopeMz(curFragMass, charge, mostAbundantIsotopeIndex);
* curObsIonMassError = (Math.Abs(mostAbuPeak.Mz - mostAbuPeakMz) / mostAbuPeakMz) * 1e6;
*
* //var curObsIonMass = Ion.GetMonoIsotopicMass(mostAbuPeak.Mz, charge, mostAbundantIsotopeIndex);
* //curObsIonMassError = (Math.Abs(curFragMass - curObsIonMass) / curFragMass) * 1e6;
* }
* ionMatch = true;
*/
var mostAbuPeakMz = Ion.GetIsotopeMz(curFragMass, charge, mostAbundantIsotopeIndex);
var curObsIonMassError = (Math.Abs(mostAbuPeak.Mz - mostAbuPeakMz) / mostAbuPeakMz) * 1e6;
stat.Count++;
stat.Intensity += Math.Min(intScore, 1.0);
stat.Corr += distCorr.Item2;
stat.Dist += distCorr.Item1;
stat.MassError += curObsIonMassError;
if (baseIonType.IsPrefix)
{
prefixHit = true;
}
else
{
suffixHit = true;
}
}
//if (!ionMatch) continue;
}
if (prefixHit)
{
preFixIonCheck[cleavageIndex] = true;
}
if (suffixHit)
{
sufFixIonCheck[cleavageIndex] = true;
}
if (prefixHit && suffixHit)
{
nComplementaryFrags++;
}
cleavageIndex++;
}
var preContCount = 0;
var sufContCount = 0;
for (var i = 0; i < preFixIonCheck.Length - 1; i++)
{
if (preFixIonCheck[i] && preFixIonCheck[i + 1])
{
preContCount++;
}
if (sufFixIonCheck[i] && sufFixIonCheck[i + 1])
{
sufContCount++;
}
}
writer.Write(activationMethodFlag);
writer.Write("\t");
writer.Write(sequence.Composition.Mass);
writer.Write("\t");
writer.Write(sequence.Count);
writer.Write("\t");
writer.Write(nComplementaryFrags);
writer.Write("\t");
writer.Write("{0}\t{1}\t{2}\t{3}\t{4}\t{5}", prefixStat.Count, preContCount, prefixStat.Intensity, prefixStat.Corr, prefixStat.Dist, prefixStat.MassError);
writer.Write("\t");
writer.Write("{0}\t{1}\t{2}\t{3}\t{4}\t{5}", suffixStat.Count, sufContCount, suffixStat.Intensity, suffixStat.Corr, suffixStat.Dist, suffixStat.MassError);
writer.Write("\n");
}