private void WriteResultsToFile(SortedSet <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\tIcScore"
);
for (var scanNum = _run.MinLcScan; scanNum <= _run.MaxLcScan; scanNum++)
{
if (matches[scanNum] == null)
{
continue;
}
foreach (var match in matches[scanNum].Reverse())
{
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 scores = _bottomUpScorer.GetScores(match, ion.Composition, ion.Charge, scanNum);
if (ion == null)
{
Console.WriteLine(@"Null ion!");
}
if (scores == null)
{
Console.WriteLine(@"Null scores");
}
// 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}",
scanNum,
match.Pre,
sequence, // Sequence
match.Post,
scores.Modifications, // Modifications
ion.Composition, // Composition
proteinName, // ProteinName
database.GetProteinDescription(match.Offset), // ProteinDescription
protLength, // ProteinLength
start, // Start
end, // End
ion.Charge, // precursorCharge
StringUtilities.DblToString(ion.GetMostAbundantIsotopeMz(), 9, true), // MostAbundantIsotopeMz
StringUtilities.DblToString(ion.Composition.Mass, 9, true), // Mass
match.Score,
scores.Score // Score (re-scored)
);
}
}
}
}
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 OutputMergedResult(TextWriter writer, TsvFileParser parser, FastaDatabase fastaDb)
{
var scoreColumn = parser.GetData("#MatchedFragments") ?? parser.GetData("Score");
var qValColumn = parser.GetData("QValue");
for (var i = 0; i < parser.NumData; i++)
{
var sequence = parser.GetData("Sequence")[i];
var scanNum = int.Parse(parser.GetData("Scan")[i]);
var mass = double.Parse(parser.GetData("Mass")[i]);
var protName = parser.GetData("ProteinName")[i];
var protDesc = fastaDb.GetProteinDescription(protName);
var firstResId = int.Parse(parser.GetData("Start")[i]);
var lastResId = int.Parse(parser.GetData("End")[i]);
var score = double.Parse(scoreColumn[i]);
var mod = parser.GetData("Modifications")[i];
var qvalue = (qValColumn != null) ? qValColumn[i] : "0";
writer.Write(scanNum);
writer.Write("\t");
writer.Write(sequence);
writer.Write("\t");
writer.Write(mod);
writer.Write("\t");
writer.Write(mass);
writer.Write("\t");
writer.Write(protName);
writer.Write("\t");
writer.Write(protDesc);
writer.Write("\t");
writer.Write(firstResId);
writer.Write("\t");
writer.Write(lastResId);
writer.Write("\t");
writer.Write(score);
writer.Write("\t");
writer.Write(qvalue);
writer.Write("\n");
}
}