private static string GetLocalizedString(LocalizedHit hit)
{
StringBuilder sb = new StringBuilder();
IMass[] mods = hit.LocalizedIsoform.GetModifications();
bool first = false;
for (int resNumber = 1; resNumber < mods.Length - 1; resNumber++)
{
IMass mod = mods[resNumber];
if (mod == null)
{
continue;
}
ModificationCollection modCollection = mod as ModificationCollection;
if (modCollection != null)
{
IMass trueMod = mod;
bool passes = false;
foreach (IMass mod2 in modCollection)
{
if (Lotor.QuantifiedModifications.Contains(mod2))
{
trueMod = mod2;
passes = true;
break;
}
}
if (!passes)
{
continue;
}
mod = trueMod;
}
else
{
if (Lotor.FixedModifications.Contains(mod) || !Lotor.QuantifiedModifications.Contains(mod))
{
continue;
}
}
int fullResidueNumber = hit.PSM.StartResidue + resNumber - 1;
char res = hit.LocalizedIsoform.Sequence[resNumber - 1];
if (first)
{
sb.Append(" & ");
}
sb.Append(res);
sb.Append(fullResidueNumber);
sb.Append("[" + mod + "]");
first = true;
}
return(sb.ToString());
}
public void AddHit(LocalizedHit hit)
{
string localizedString = GetLocalizedString(hit);
List<LocalizedHit> hits = null;
if (Hits.TryGetValue(localizedString, out hits))
{
hits.Add(hit);
}
else
{
hits = new List<LocalizedHit>();
hits.Add(hit);
Hits.Add(localizedString, hits);
}
}
public void AddHit(LocalizedHit hit)
{
string localizedString = GetLocalizedString(hit);
List <LocalizedHit> hits = null;
if (Hits.TryGetValue(localizedString, out hits))
{
hits.Add(hit);
}
else
{
hits = new List <LocalizedHit>();
hits.Add(hit);
Hits.Add(localizedString, hits);
}
}
private static string GetLocalizedString(LocalizedHit hit)
{
StringBuilder sb = new StringBuilder();
IMass[] mods = hit.LocalizedIsoform.GetModifications();
bool first = false;
for(int resNumber = 1; resNumber < mods.Length-1; resNumber++)
{
IMass mod = mods[resNumber];
if (mod == null)
continue;
ModificationCollection modCollection = mod as ModificationCollection;
if (modCollection != null)
{
IMass trueMod = mod;
bool passes = false;
foreach (IMass mod2 in modCollection)
{
if (Lotor.QuantifiedModifications.Contains(mod2))
{
trueMod = mod2;
passes = true;
break;
}
}
if (!passes)
continue;
mod = trueMod;
}
else
{
if (Lotor.FixedModifications.Contains(mod) || !Lotor.QuantifiedModifications.Contains(mod))
continue;
}
int fullResidueNumber = hit.PSM.StartResidue + resNumber - 1;
char res = hit.LocalizedIsoform.Sequence[resNumber-1];
if (first)
{
sb.Append(" & ");
}
sb.Append(res);
sb.Append(fullResidueNumber);
sb.Append("[" + mod + "]");
first = true;
}
return sb.ToString();
}
private List<LocalizedHit> CalculateBestIsoforms(List<PSM> psms, FragmentTypes fragType, Tolerance prod_tolerance, double productThreshold, bool phosphoNeutralLosses)
{
Log("Localizing Best Isoforms...");
int totalisofromscount = 0;
int count = 0;
int psm_count = 0;
int localized_psm = 0;
List<LocalizedHit> hits = new List<LocalizedHit>();
foreach (PSM psm in psms)
{
psm_count++;
count++;
// Generate all the isoforms for the PSM
int isoformCount = psm.GenerateIsoforms(_ignoreCTerminal);
if(isoformCount == 0)
continue;
totalisofromscount += isoformCount;
// Calculate the probability of success for random matches
//double pvalue = GetPValue(psm, prod_tolerance, productThreshold);
// Match all the isoforms to the spectrum and log the results
psm.MatchIsofroms(fragType, prod_tolerance, productThreshold, phosphoNeutralLosses, 1);
// Perform the localization for all combinations of isoforms
//double[,] res = psm.Calc(pvalue);
Tuple<int, int, double> scores = LocalizedIsoformSimple(psm);
// Check if the localization is above some threshold
//Tuple<int, int, double> scores = LocalizedIsoform(res);
int bestIsoform = scores.Item1;
int secondBestIsoform = scores.Item2;
double ascore = scores.Item3;
//int numSDFs = psm.NumSiteDeterminingFragments[bestIsoform, secondBestIsoform];
List<PeptideIsoform> isoforms = psm.PeptideIsoforms.ToList();
//LocalizedHit hit = new LocalizedHit(psm, isoforms[bestIsoform], isoforms[secondBestIsoform], numSDFs,
// psm.BestSiteDeterminingFragments[bestIsoform, secondBestIsoform],
// psm.BestSiteDeterminingFragments[secondBestIsoform, bestIsoform], pvalue, ascore);
LocalizedHit hit = new LocalizedHit(psm, isoforms[bestIsoform], isoforms[secondBestIsoform], 0,
0,
0, 0, ascore);
hits.Add(hit);
if (hit.IsLocalized)
{
localized_psm++;
}
psm.PeptideIsoforms.Clear();
// Progress Bar Stuff
if (count <= 50)
continue;
count = 0;
ProgressUpdate((double)psm_count / psms.Count);
}
ProgressUpdate(1.0);
Log(string.Format("Total Number of Possible Isoforms Considered: {0:N0}",totalisofromscount));
Log(string.Format("Total Number of PSMs Considered: {0:N0}", psm_count));
Log(string.Format("Total Number of PSMs Localized: {0:N0} ({1:00.00}%)", localized_psm, localized_psm * 100 / psm_count));
ProgressUpdate(0);
return hits;
}
private List <LocalizedHit> CalculateBestIsoforms(List <PSM> psms, FragmentTypes fragType, Tolerance prod_tolerance, double productThreshold, bool phosphoNeutralLosses)
{
Log("Localizing Best Isoforms...");
int totalisofromscount = 0;
int count = 0;
int psm_count = 0;
int localized_psm = 0;
List <LocalizedHit> hits = new List <LocalizedHit>();
foreach (PSM psm in psms)
{
psm_count++;
count++;
// Generate all the isoforms for the PSM
int isoformCount = psm.GenerateIsoforms(_ignoreCTerminal);
if (isoformCount == 0)
{
continue;
}
totalisofromscount += isoformCount;
// Calculate the probability of success for random matches
//double pvalue = GetPValue(psm, prod_tolerance, productThreshold);
// Match all the isoforms to the spectrum and log the results
psm.MatchIsofroms(fragType, prod_tolerance, productThreshold, phosphoNeutralLosses, 1);
// Perform the localization for all combinations of isoforms
//double[,] res = psm.Calc(pvalue);
Tuple <int, int, double> scores = LocalizedIsoformSimple(psm);
// Check if the localization is above some threshold
//Tuple<int, int, double> scores = LocalizedIsoform(res);
int bestIsoform = scores.Item1;
int secondBestIsoform = scores.Item2;
double ascore = scores.Item3;
//int numSDFs = psm.NumSiteDeterminingFragments[bestIsoform, secondBestIsoform];
List <PeptideIsoform> isoforms = psm.PeptideIsoforms.ToList();
//LocalizedHit hit = new LocalizedHit(psm, isoforms[bestIsoform], isoforms[secondBestIsoform], numSDFs,
// psm.BestSiteDeterminingFragments[bestIsoform, secondBestIsoform],
// psm.BestSiteDeterminingFragments[secondBestIsoform, bestIsoform], pvalue, ascore);
LocalizedHit hit = new LocalizedHit(psm, isoforms[bestIsoform], isoforms[secondBestIsoform], 0,
0,
0, 0, ascore);
hits.Add(hit);
if (hit.IsLocalized)
{
localized_psm++;
}
psm.PeptideIsoforms.Clear();
// Progress Bar Stuff
if (count <= 50)
{
continue;
}
count = 0;
ProgressUpdate((double)psm_count / psms.Count);
}
ProgressUpdate(1.0);
Log(string.Format("Total Number of Possible Isoforms Considered: {0:N0}", totalisofromscount));
Log(string.Format("Total Number of PSMs Considered: {0:N0}", psm_count));
Log(string.Format("Total Number of PSMs Localized: {0:N0} ({1:00.00}%)", localized_psm, localized_psm * 100 / psm_count));
ProgressUpdate(0);
return(hits);
}
private List <Protein> CompileResults(List <LocalizedHit> hits, string csvFile, string outputDirectory, bool breakProteinsApart = false)
{
Dictionary <string, LocalizedHit> hitsdict = new Dictionary <string, LocalizedHit>();
// Group all the localized Hits into proteins
Dictionary <string, Protein> proteins = new Dictionary <string, Protein>();
foreach (LocalizedHit hit in hits)
{
hitsdict.Add(hit.PSM.Filename, hit);
string defline = hit.PSM.Defline;
if (breakProteinsApart)
{
string[] groups = hit.PSM.ProteinGroup.Split('|');
foreach (string group in groups)
{
Protein prot;
if (!proteins.TryGetValue(group, out prot))
{
prot = new Protein(group, defline);
proteins.Add(group, prot);
}
prot.AddHit(hit);
}
}
else
{
Protein prot;
if (!proteins.TryGetValue(hit.PSM.ProteinGroup, out prot))
{
prot = new Protein(hit.PSM.ProteinGroup, defline);
proteins.Add(hit.PSM.ProteinGroup, prot);
}
prot.AddHit(hit);
}
}
using (StreamWriter writer = new StreamWriter(Path.Combine(outputDirectory, Path.GetFileNameWithoutExtension(csvFile) + "_all.csv")),
localizedWriter = new StreamWriter(Path.Combine(outputDirectory, Path.GetFileNameWithoutExtension(csvFile) + "_localized.csv")))
{
using (CsvReader reader = new CsvReader(new StreamReader(csvFile), true))
{
LocalizedHit hit = null;
headerInfo = reader.GetFieldHeaders();
bool tqFound = false;
for (int i = 0; i < reader.FieldCount; i++)
{
if (headerInfo[i].EndsWith("NL)"))
{
if (!tqFound)
{
FirstQuantColumn = i;
tqFound = true;
}
}
if (headerInfo[i] == "Channels Detected")
{
LastQuantColumn = i - 1;
}
}
string header = string.Join(",", headerInfo) + ",# Isoforms,# of Considered Fragments,Localized?,Delta Score,Best Isoform,Spectral Matches,% TIC,Second Best Isoform,Second Spectral Matches,Second % TIC";
writer.WriteLine(header);
localizedWriter.WriteLine(header);
while (reader.ReadNextRecord())
{
string mods = reader["Mods"];
if (string.IsNullOrEmpty(mods))
{
continue;
}
List <Modification> variableMods = OmssaModification.ParseModificationLine(mods).Select(item => item.Item1).OfType <Modification>().ToList();
// Only keep things with quantified Modifications
if (!variableMods.Any(mod => QuantifiedModifications.Contains(mod)))
{
continue;
}
string filename = reader["Filename/id"];
if (!hitsdict.TryGetValue(filename, out hit))
{
continue;
}
string[] data = new string[reader.FieldCount];
reader.CopyCurrentRecordTo(data);
hit.omssapsm = data;
StringBuilder sb = new StringBuilder();
foreach (string datum in data)
{
if (datum.Contains(','))
{
sb.Append("\"");
sb.Append(datum);
sb.Append("\"");
}
else
{
sb.Append(datum);
}
sb.Append(',');
}
sb.Append(hit.PSM.Isoforms);
sb.Append(',');
sb.Append(hit.LocalizedIsoform.Fragments.Count);
sb.Append(',');
sb.Append(hit.IsLocalized);
sb.Append(',');
sb.Append(hit.MatchDifference);
sb.Append(',');
sb.Append(hit.LocalizedIsoform.SequenceWithModifications);
sb.Append(',');
sb.Append(hit.LocalizedIsoform.SpectralMatch.Matches);
sb.Append(',');
sb.Append(hit.LocalizedIsoform.SpectralMatch.PercentTIC);
if (hit.PSM.Isoforms > 1)
{
//sb.Append(',');
//sb.Append(hit.BestPeptideSDFCount);
sb.Append(',');
sb.Append(hit.SecondBestPeptideIsoform.SequenceWithModifications);
sb.Append(',');
sb.Append(hit.SecondBestPeptideIsoform.SpectralMatch.Matches);
sb.Append(',');
sb.Append(hit.SecondBestPeptideIsoform.SpectralMatch.PercentTIC);
//sb.Append(',');
//sb.Append(hit.SecondBestPeptideSDFCount);
}
if (hit.IsLocalized)
{
localizedWriter.WriteLine(sb.ToString());
}
writer.WriteLine(sb.ToString());
}
}
}
return(proteins.Values.ToList());
}
