public static double GetPValue(PSM psm, Tolerance prod_tolerance, double cutoff)
{
// Get the width of the product ion tolerance at the isolation mz;
double productToleranceWidth = prod_tolerance.GetRange(psm.IsolationMZ).Width;
double cutoffThreshold = psm.Spectrum.GetBasePeakIntensity()*cutoff;
return Math.Min(1.0, psm.Spectrum.Count(peak => peak.Intensity >= cutoffThreshold)*2*productToleranceWidth/psm.ScanWidth);
}
public static double GetPValue(PSM psm, Tolerance prod_tolerance, double cutoff)
{
// Get the width of the product ion tolerance at the isolation mz;
double productToleranceWidth = prod_tolerance.GetRange(psm.IsolationMZ).Width;
double cutoffThreshold = psm.Spectrum.GetBasePeakIntensity() * cutoff;
return(Math.Min(1.0, psm.Spectrum.Count(peak => peak.Intensity >= cutoffThreshold) * 2 * productToleranceWidth / psm.ScanWidth));
}
private Tuple <int, int, double> LocalizedIsoformSimple(PSM psm)
{
int length = psm.Isoforms;
List <PeptideIsoform> isoforms = psm.PeptideIsoforms.ToList();
if (length == 1)
{
return(new Tuple <int, int, double>(0, 0, double.PositiveInfinity));
}
int biggesti = 0;
int bestI = 0;
for (int i = 0; i < length; i++)
{
var isoform = isoforms[i];
if (isoform.SpectralMatch.Matches > biggesti)
{
biggesti = isoform.SpectralMatch.Matches;
bestI = i;
}
}
int biggestj = 0;
int bestJ = bestI;
for (int j = 0; j < length; j++)
{
if (j == bestI)
{
continue;
}
var isoform = isoforms[j];
if (isoform.SpectralMatch.Matches > biggestj)
{
biggestj = isoform.SpectralMatch.Matches;
bestJ = j;
}
}
return(new Tuple <int, int, double>(bestI, bestJ, double.PositiveInfinity));
}
private Tuple<int, int, double> LocalizedIsoformSimple(PSM psm)
{
int length = psm.Isoforms;
List<PeptideIsoform> isoforms = psm.PeptideIsoforms.ToList();
if (length == 1)
return new Tuple<int, int, double>(0, 0, double.PositiveInfinity);
int biggesti = 0;
int bestI = 0;
for (int i = 0; i < length; i++)
{
var isoform = isoforms[i];
if (isoform.SpectralMatch.Matches > biggesti)
{
biggesti = isoform.SpectralMatch.Matches;
bestI = i;
}
}
int biggestj = 0;
int bestJ = bestI;
for (int j = 0; j < length; j++)
{
if (j == bestI)
continue;
var isoform = isoforms[j];
if (isoform.SpectralMatch.Matches > biggestj)
{
biggestj = isoform.SpectralMatch.Matches;
bestJ = j;
}
}
return new Tuple<int, int, double>(bestI, bestJ, double.PositiveInfinity);
}
private List<PSM> LoadAllPSMs(string csvFile, string rawFileDirectory, List<Modification> fixedMods)
{
ProgressUpdate(0.0); //force the progressbar to go into marquee mode
Log("Reading PSMs from " + csvFile);
Dictionary<string, ThermoRawFile> rawFiles =
Directory.EnumerateFiles(rawFileDirectory, "*.raw", SearchOption.AllDirectories)
.ToDictionary(Path.GetFileNameWithoutExtension, file => new ThermoRawFile(file));
_dataFiles = new HashSet<MSDataFile>();
List<PSM> psms = new List<PSM>();
int totalPsms = 0;
using (CsvReader reader = new CsvReader(new StreamReader(csvFile), true))
{
while (reader.ReadNextRecord())
{
string mods = reader["Mods"];
totalPsms++;
// Skip if there are no modifications
if (string.IsNullOrEmpty(mods))
continue;
// Convert the text mod line into a list of modification objects
List<Modification> variableMods = OmssaModification.ParseModificationLine(mods).Select(item => item.Item1).ToList();
// Only keep things with quantified Modifications
if (!variableMods.Any(mod => QuantifiedModifications.Contains(mod)))
continue;
string filename = reader["Filename/id"];
string rawname = filename.Split('.')[0];
int scanNumber = int.Parse(reader["Spectrum number"]);
PSM psm = new PSM(scanNumber, rawname);
psm.StartResidue = int.Parse(reader["Start"]);
psm.Charge = int.Parse(reader["Charge"]);
psm.BasePeptide = new Peptide(reader["Peptide"].ToUpper());
psm.Defline = reader["Defline"];
psm.ProteinGroup = reader["Best PG Name"];
psm.NumberOfSharingProteinGroups = int.Parse(reader["# of Sharing PGs"]);
psm.Filename = filename;
// Apply all the fix modifications
psm.BasePeptide.SetModifications(fixedMods);
int i = 0;
while (i < variableMods.Count)
{
if (fixedMods.Contains(variableMods[i]))
{
variableMods.RemoveAt(i);
}
else
{
i++;
}
}
// Save all the variable mod types
psm.VariabledModifications = variableMods;
psms.Add(psm);
}
}
Log(string.Format("{0:N0} PSMs were loaded....", totalPsms));
Log(string.Format("{0:N0} PSMs were kept.... ({1:F2} %)", psms.Count, 100.0 * (double)psms.Count / totalPsms));
Log("Reading Spectral Data...");
ThermoRawFile currentRawFile = null;
string currentRawFileName = null;
int counter = 0;
foreach (PSM psm in psms.OrderBy(psm => psm.RawFileName))
{
string rawfilename = psm.RawFileName;
if(!rawfilename.Equals(currentRawFileName)) {
currentRawFileName = rawfilename;
if (currentRawFile != null && currentRawFile.IsOpen)
currentRawFile.Dispose();
if (!rawFiles.TryGetValue(rawfilename, out currentRawFile))
{
throw new NullReferenceException(string.Format("Raw File: {0}.raw was not found! Aborting.", rawfilename));
}
currentRawFile.Open();
}
psm.SetRawFile(currentRawFile);
counter++;
if(counter % 25 == 0)
{
ProgressUpdate((double)counter / psms.Count);
}
}
return psms;
}
private List <PSM> LoadAllPSMs(string csvFile, string rawFileDirectory, List <Modification> fixedMods)
{
ProgressUpdate(0.0); //force the progressbar to go into marquee mode
Log("Reading PSMs from " + csvFile);
Dictionary <string, ThermoRawFile> rawFiles =
Directory.EnumerateFiles(rawFileDirectory, "*.raw", SearchOption.AllDirectories)
.ToDictionary(Path.GetFileNameWithoutExtension, file => new ThermoRawFile(file));
_dataFiles = new HashSet <MSDataFile>();
List <PSM> psms = new List <PSM>();
int totalPsms = 0;
using (CsvReader reader = new CsvReader(new StreamReader(csvFile), true))
{
while (reader.ReadNextRecord())
{
string mods = reader["Mods"];
totalPsms++;
// Skip if there are no modifications
if (string.IsNullOrEmpty(mods))
{
continue;
}
// Convert the text mod line into a list of modification objects
List <Modification> variableMods = OmssaModification.ParseModificationLine(mods).Select(item => item.Item1).ToList();
// Only keep things with quantified Modifications
if (!variableMods.Any(mod => QuantifiedModifications.Contains(mod)))
{
continue;
}
string filename = reader["Filename/id"];
string rawname = filename.Split('.')[0];
int scanNumber = int.Parse(reader["Spectrum number"]);
PSM psm = new PSM(scanNumber, rawname);
psm.StartResidue = int.Parse(reader["Start"]);
psm.Charge = int.Parse(reader["Charge"]);
psm.BasePeptide = new Peptide(reader["Peptide"].ToUpper());
psm.Defline = reader["Defline"];
psm.ProteinGroup = reader["Best PG Name"];
psm.NumberOfSharingProteinGroups = int.Parse(reader["# of Sharing PGs"]);
psm.Filename = filename;
// Apply all the fix modifications
psm.BasePeptide.SetModifications(fixedMods);
int i = 0;
while (i < variableMods.Count)
{
if (fixedMods.Contains(variableMods[i]))
{
variableMods.RemoveAt(i);
}
else
{
i++;
}
}
// Save all the variable mod types
psm.VariabledModifications = variableMods;
psms.Add(psm);
}
}
Log(string.Format("{0:N0} PSMs were loaded....", totalPsms));
Log(string.Format("{0:N0} PSMs were kept.... ({1:F2} %)", psms.Count, 100.0 * (double)psms.Count / totalPsms));
Log("Reading Spectral Data...");
ThermoRawFile currentRawFile = null;
string currentRawFileName = null;
int counter = 0;
foreach (PSM psm in psms.OrderBy(psm => psm.RawFileName))
{
string rawfilename = psm.RawFileName;
if (!rawfilename.Equals(currentRawFileName))
{
currentRawFileName = rawfilename;
if (currentRawFile != null && currentRawFile.IsOpen)
{
currentRawFile.Dispose();
}
if (!rawFiles.TryGetValue(rawfilename, out currentRawFile))
{
throw new NullReferenceException(string.Format("Raw File: {0}.raw was not found! Aborting.", rawfilename));
}
currentRawFile.Open();
}
psm.SetRawFile(currentRawFile);
counter++;
if (counter % 25 == 0)
{
ProgressUpdate((double)counter / psms.Count);
}
}
return(psms);
}
