private static Dictionary <string, Protease> LoadProteaseDictionary(string proteasesLocation)
{
Dictionary <string, Protease> dict = new Dictionary <string, Protease>();
using (StreamReader proteases = new StreamReader(proteasesLocation))
{
proteases.ReadLine();
while (proteases.Peek() != -1)
{
string line = proteases.ReadLine();
string[] fields = line.Split('\t');
string name = fields[0];
string[] sequences_inducing_cleavage = fields[1].Split(new char[] { ',' }, StringSplitOptions.RemoveEmptyEntries);
string[] sequences_preventing_cleavage = fields[2].Split(new char[] { ',' }, StringSplitOptions.RemoveEmptyEntries);
var cleavage_terminus = (TerminusType)Enum.Parse(typeof(TerminusType), fields[3], true);
var cleavage_specificity = (CleavageSpecificity)Enum.Parse(typeof(CleavageSpecificity), fields[4], true);
string psi_ms_accession_number = fields[5];
string psi_ms_name = fields[6];
string site_regexp = fields[7];
var protease = new Protease(name, sequences_inducing_cleavage, sequences_preventing_cleavage, cleavage_terminus, cleavage_specificity, psi_ms_accession_number, psi_ms_name, site_regexp);
dict.Add(protease.Name, protease);
}
}
return(dict);
}
/// <summary>
/// Gets peptides for semispecific digestion of a protein
/// </summary>
/// <param name="protein"></param>
/// <returns></returns>
public IEnumerable <PeptideWithSetModifications> SemiSpecificDigestion(Protein protein)
{
List <Peptide> intervals = new List <Peptide>();
List <int> oneBasedIndicesToCleaveAfter = Protease.GetDigestionSiteIndices(protein.BaseSequence);
for (int i = 0; i < oneBasedIndicesToCleaveAfter.Count - MaximumMissedCleavages - 1; i++)
{
if (Protease.Retain(i, InitiatorMethionineBehavior, protein[0]) &&
Protease.OkayLength(oneBasedIndicesToCleaveAfter[i + MaximumMissedCleavages + 1] - oneBasedIndicesToCleaveAfter[i], MinPeptidesLength, MaxPeptidesLength))
{
intervals.Add(new Peptide(protein, oneBasedIndicesToCleaveAfter[i] + 1, oneBasedIndicesToCleaveAfter[i + MaximumMissedCleavages + 1],
oneBasedIndicesToCleaveAfter[i + MaximumMissedCleavages + 1] - oneBasedIndicesToCleaveAfter[i], "semi"));
}
if (Protease.Cleave(i, InitiatorMethionineBehavior, protein[0]) &&
Protease.OkayLength(oneBasedIndicesToCleaveAfter[i + MaximumMissedCleavages + 1] - 1, MinPeptidesLength, MaxPeptidesLength))
{
intervals.Add(new Peptide(protein, 2, oneBasedIndicesToCleaveAfter[i + MaximumMissedCleavages + 1],
oneBasedIndicesToCleaveAfter[i + MaximumMissedCleavages + 1] - 1, "semi:M cleaved"));
}
}
int lastIndex = oneBasedIndicesToCleaveAfter.Count - 1;
int maxIndex = MaximumMissedCleavages < lastIndex ? MaximumMissedCleavages : lastIndex;
for (int i = 1; i <= maxIndex; i++)
{
if (DigestionParams.TerminusTypeSemiProtease == TerminusType.N) //tricky, it's N because we want the extra peptide at the C terminus |_
{
if (Protease.OkayLength(oneBasedIndicesToCleaveAfter[lastIndex] - oneBasedIndicesToCleaveAfter[lastIndex - i], MinPeptidesLength, MaxPeptidesLength))
{
intervals.Add(new Peptide(protein, oneBasedIndicesToCleaveAfter[lastIndex - i] + 1, oneBasedIndicesToCleaveAfter[lastIndex],
oneBasedIndicesToCleaveAfter[lastIndex] - oneBasedIndicesToCleaveAfter[lastIndex - i], "semiN"));
}
}
else //TerminusType.C
{
if (Protease.OkayLength(oneBasedIndicesToCleaveAfter[i] - oneBasedIndicesToCleaveAfter[0], MinPeptidesLength, MaxPeptidesLength))
{
intervals.Add(new Peptide(protein, oneBasedIndicesToCleaveAfter[0] + 1, oneBasedIndicesToCleaveAfter[i],
oneBasedIndicesToCleaveAfter[i] - oneBasedIndicesToCleaveAfter[0], "semiC"));
}
}
}
// Also digest using the proteolysis product start/end indices
intervals.AddRange(
protein.ProteolysisProducts
.Where(proteolysisProduct => proteolysisProduct.OneBasedBeginPosition != 1 || proteolysisProduct.OneBasedEndPosition != protein.Length)
.Select(proteolysisProduct => new Peptide(protein, proteolysisProduct.OneBasedBeginPosition.Value, proteolysisProduct.OneBasedEndPosition.Value,
0, proteolysisProduct.Type + " start")));
return(intervals.SelectMany(peptide => peptide.GetModifiedPeptides(AllKnownFixedModifications, DigestionParams, VariableModifications)));
}
/// <summary>
/// Gets peptides for specific protease digestion of a protein
/// </summary>
/// <param name="protein"></param>
/// <returns></returns>
public IEnumerable <PeptideWithSetModifications> Digestion(Protein protein)
{
var intervals = Protease.GetDigestionIntervals(protein, MaximumMissedCleavages, InitiatorMethionineBehavior, MinPeptidesLength, MaxPeptidesLength);
return(intervals.SelectMany(peptide => peptide.GetModifiedPeptides(AllKnownFixedModifications, DigestionParams, VariableModifications)));
}
private List <ProteinGroup> ApplyProteinParsimony()
{
//if dictionary is empty return an empty list of protein groups
if (!CompactPeptideToProteinPeptideMatching.Values.Any())
{
return(new List <ProteinGroup>());
}
// digesting an XML database results in a non-mod-agnostic digestion; need to fix this if mod-agnostic parsimony enabled
if (!TreatModPeptidesAsDifferentPeptides)//user want modified and unmodified peptides treated the same
{
Dictionary <string, HashSet <PeptideWithSetModifications> > baseSeqToProteinMatch = new Dictionary <string, HashSet <PeptideWithSetModifications> >();
// dictionary where string key is the base sequence and the HashSet is all PeptidesWithSetModificatiosn with the same sequence
// can access which protein these matching peptides came from through the PeptideWithSetModifications object
foreach (var peptide in CompactPeptideToProteinPeptideMatching.SelectMany(b => b.Value))
{
if (baseSeqToProteinMatch.TryGetValue(peptide.BaseSequence, out HashSet <PeptideWithSetModifications> value))
{
value.Add(peptide);
}
else
{
baseSeqToProteinMatch[peptide.BaseSequence] = new HashSet <PeptideWithSetModifications> {
peptide
};
}
}
var blah = new Dictionary <PeptideWithSetModifications, List <CompactPeptideBase> >();
// where to store results
foreach (var pep in CompactPeptideToProteinPeptideMatching)
{
foreach (var pepWithSetMods in pep.Value)
{
if (blah.TryGetValue(pepWithSetMods, out List <CompactPeptideBase> list))
{
list.Add(pep.Key);
}
else
{
blah.Add(pepWithSetMods, new List <CompactPeptideBase> {
pep.Key
});
}
}
}
foreach (var baseSequence in baseSeqToProteinMatch)
{
if (baseSequence.Value.Count > 1 && baseSequence.Value.Any(p => p.NumMods > 0))
{
// list of proteins along with start/end residue in protein and the # missed cleavages
var peptideInProteinInfo = new List <Tuple <Protein, DigestionParams, int, int, int> >();
foreach (var peptide in baseSequence.Value)
{
peptideInProteinInfo.Add(new Tuple <Protein, DigestionParams, int, int, int>(peptide.Protein, peptide.DigestionParams, peptide.OneBasedStartResidueInProtein, peptide.OneBasedEndResidueInProtein, (int)peptide.MissedCleavages));
}
foreach (var peptide in baseSequence.Value)
{
foreach (var proteinInfo in peptideInProteinInfo)
{
var pep = new PeptideWithSetModifications(proteinInfo.Item1, proteinInfo.Item2, proteinInfo.Item3, proteinInfo.Item4, peptide.PeptideDescription, proteinInfo.Item5, peptide.AllModsOneIsNterminus, peptide.NumFixedMods);
foreach (var compactPeptide in blah[peptide])
{
CompactPeptideToProteinPeptideMatching[compactPeptide].Add(pep);
}
}
}
}
}
}
var proteinToPeptidesMatching = new Dictionary <Protein, HashSet <CompactPeptideBase> >();
var parsimonyProteinList = new Dictionary <Protein, HashSet <CompactPeptideBase> >();
var proteinsWithUniquePeptides = new Dictionary <Protein, HashSet <PeptideWithSetModifications> >();
// peptide matched to fullseq (used depending on user preference)
var compactPeptideToFullSeqMatch = CompactPeptideToProteinPeptideMatching.ToDictionary(x => x.Key, x => x.Value.First().Sequence);
foreach (var kvp in CompactPeptideToProteinPeptideMatching)
{
HashSet <Protein> proteinsAssociatedWithThisPeptide = new HashSet <Protein>(kvp.Value.Select(p => p.Protein));
if (proteinsAssociatedWithThisPeptide.Count == 1)
{
if (!proteinsWithUniquePeptides.TryGetValue(kvp.Value.First().Protein, out HashSet <PeptideWithSetModifications> peptides))
{
proteinsWithUniquePeptides.Add(kvp.Value.First().Protein, new HashSet <PeptideWithSetModifications>(kvp.Value));
}
else
{
peptides.UnionWith(kvp.Value);
}
}
// multiprotease parsimony is "weird" because a peptide sequence can be shared between
// two proteins but technically be a "unique" peptide because it is unique in that protease digestion
// this code marks these types of peptides as unique
else
{
foreach (var peptide in kvp.Value)
{
Protease protease = peptide.DigestionParams.Protease;
int sameProteaseCount = kvp.Value.Count(v => v.DigestionParams.Protease == protease);
if (sameProteaseCount == 1)
{
if (!proteinsWithUniquePeptides.TryGetValue(peptide.Protein, out HashSet <PeptideWithSetModifications> peps))
{
proteinsWithUniquePeptides.Add(peptide.Protein, new HashSet <PeptideWithSetModifications> {
peptide
});
}
else
{
peps.UnionWith(kvp.Value);
}
}
}
}
// if a peptide is associated with a decoy protein, remove all target protein associations with the peptide
if (kvp.Value.Any(p => p.Protein.IsDecoy))
{
kvp.Value.RemoveWhere(p => !p.Protein.IsDecoy);
}
// if a peptide is associated with a contaminant protein, remove all target protein associations with the peptide
if (kvp.Value.Any(p => p.Protein.IsContaminant))
{
kvp.Value.RemoveWhere(p => !p.Protein.IsContaminant);
}
}
// makes dictionary with proteins as keys and list of associated peptides as the value (makes parsimony algo easier)
foreach (var kvp in CompactPeptideToProteinPeptideMatching)
{
foreach (var peptide in kvp.Value)
{
if (!proteinToPeptidesMatching.TryGetValue(peptide.Protein, out HashSet <CompactPeptideBase> peptides))
{
proteinToPeptidesMatching.Add(peptide.Protein, new HashSet <CompactPeptideBase>()
{
kvp.Key
});
}
else
{
peptides.Add(kvp.Key);
}
}
}
// build protein list for each peptide before parsimony has been applied
var peptideSeqProteinListMatch = new Dictionary <string, HashSet <Protein> >();
foreach (var kvp in proteinToPeptidesMatching)
{
foreach (var peptide in kvp.Value)
{
string pepSequence;
if (!TreatModPeptidesAsDifferentPeptides)
{
string nTerminalMasses = peptide.NTerminalMasses == null ? "" : string.Join("", peptide.NTerminalMasses.Select(b => b.ToString(CultureInfo.InvariantCulture)));
string cTerminalMasses = peptide.CTerminalMasses == null ? "" : string.Join("", peptide.CTerminalMasses.Select(b => b.ToString(CultureInfo.InvariantCulture)));
pepSequence = nTerminalMasses + cTerminalMasses + peptide.MonoisotopicMassIncludingFixedMods.ToString(CultureInfo.InvariantCulture);
}
else
{
pepSequence = compactPeptideToFullSeqMatch[peptide];
}
if (!peptideSeqProteinListMatch.TryGetValue(pepSequence, out HashSet <Protein> proteinListHere))
{
peptideSeqProteinListMatch.Add(pepSequence, new HashSet <Protein>()
{
kvp.Key
});
}
else
{
proteinListHere.Add(kvp.Key);
}
}
}
// dictionary associates proteins w/ unused base seqs (list will shrink over time)
var algDictionary = new Dictionary <Protein, HashSet <string> >();
foreach (var kvp in peptideSeqProteinListMatch)
{
foreach (var protein in kvp.Value)
{
if (algDictionary.TryGetValue(protein, out HashSet <string> newPeptideBaseSeqs))
{
newPeptideBaseSeqs.Add(kvp.Key);
}
else
{
algDictionary.Add(protein, new HashSet <string> {
kvp.Key
});
}
}
}
// dictionary associates proteins w/ unused base seqs (list will NOT shrink over time)
var proteinToPepSeqMatch = algDictionary.ToDictionary(x => x.Key, x => x.Value);
// *** main parsimony loop
bool uniquePeptidesLeft = proteinsWithUniquePeptides.Any();
int numNewSeqs = algDictionary.Max(p => p.Value.Count);
while (numNewSeqs != 0)
{
var possibleBestProteinList = new List <KeyValuePair <Protein, HashSet <string> > >();
if (uniquePeptidesLeft)
{
var proteinsWithUniquePeptidesLeft = algDictionary.Where(p => proteinsWithUniquePeptides.ContainsKey(p.Key));
if (proteinsWithUniquePeptidesLeft.Any())
{
possibleBestProteinList.Add(proteinsWithUniquePeptidesLeft.First());
}
else
{
uniquePeptidesLeft = false;
}
}
// gets list of proteins with the most unaccounted-for peptide base sequences
if (!uniquePeptidesLeft)
{
possibleBestProteinList = algDictionary.Where(p => p.Value.Count == numNewSeqs).ToList();
}
Protein bestProtein = possibleBestProteinList.First().Key;
HashSet <string> newSeqs = new HashSet <string>(algDictionary[bestProtein]);
// may need to select different protein
if (possibleBestProteinList.Count > 1)
{
var proteinsWithTheseBaseSeqs = new HashSet <Protein>();
foreach (var kvp in possibleBestProteinList)
{
if (newSeqs.IsSubsetOf(kvp.Value))
{
proteinsWithTheseBaseSeqs.Add(kvp.Key);
}
}
if (proteinsWithTheseBaseSeqs.Count > 1)
{
var proteinsOrderedByTotalPeptideCount = new Dictionary <Protein, HashSet <string> >();
foreach (var protein in proteinsWithTheseBaseSeqs)
{
proteinsOrderedByTotalPeptideCount.Add(protein, proteinToPepSeqMatch[protein]);
}
bestProtein = proteinsOrderedByTotalPeptideCount.OrderByDescending(kvp => kvp.Value.Count).First().Key;
}
}
parsimonyProteinList.Add(bestProtein, proteinToPeptidesMatching[bestProtein]);
// remove used peptides from their proteins
foreach (var newBaseSeq in newSeqs)
{
HashSet <Protein> proteinsWithThisPeptide = peptideSeqProteinListMatch[newBaseSeq];
foreach (var protein in proteinsWithThisPeptide)
{
algDictionary[protein].Remove(newBaseSeq);
}
}
algDictionary.Remove(bestProtein);
numNewSeqs = algDictionary.Any() ? algDictionary.Max(p => p.Value.Count) : 0;
}
// *** done with parsimony
// add indistinguishable proteins
var proteinsGroupedByNumPeptides = proteinToPeptidesMatching.GroupBy(p => p.Value.Count);
var parsimonyProteinsGroupedByNumPeptides = parsimonyProteinList.GroupBy(p => p.Value.Count);
var indistinguishableProteins = new ConcurrentDictionary <Protein, HashSet <CompactPeptideBase> >();
foreach (var group in proteinsGroupedByNumPeptides)
{
var parsimonyProteinsWithSameNumPeptides = parsimonyProteinsGroupedByNumPeptides.FirstOrDefault(p => p.Key == group.Key);
var list = group.ToList();
if (parsimonyProteinsWithSameNumPeptides != null)
{
Parallel.ForEach(Partitioner.Create(0, list.Count),
new ParallelOptions {
MaxDegreeOfParallelism = commonParameters.MaxThreadsToUsePerFile
},
(range, loopState) =>
{
for (int i = range.Item1; i < range.Item2; i++)
{
foreach (var parsimonyProteinWithThisNumPeptides in parsimonyProteinsWithSameNumPeptides)
{
if (parsimonyProteinWithThisNumPeptides.Key != list[i].Key &&
proteinToPeptidesMatching[parsimonyProteinWithThisNumPeptides.Key].SetEquals(proteinToPeptidesMatching[list[i].Key]))
{
indistinguishableProteins.GetOrAdd(list[i].Key, proteinToPeptidesMatching[list[i].Key]);
}
}
}
}
);
}
}
foreach (var protein in indistinguishableProteins)
{
if (!parsimonyProteinList.ContainsKey(protein.Key))
{
parsimonyProteinList.Add(protein.Key, protein.Value);
}
}
// multiprotease parsimony:
// this code is a workaround to add back proteins to the parsimonious list that were removed
// because unique peptides were mistaken for shared peptides. see line 139 for more info
if (ListOfDigestionParams.Select(v => v.Protease).Distinct().Count() > 1)
{
HashSet <Protein> parsimonyProteinSet = new HashSet <Protein>(parsimonyProteinList.Keys);
// add back in proteins that contain unique peptides
foreach (var prot in proteinsWithUniquePeptides)
{
if (!parsimonyProteinSet.Contains(prot.Key))
{
parsimonyProteinList.Add(prot.Key, proteinToPeptidesMatching[prot.Key]);
}
}
}
foreach (var kvp in CompactPeptideToProteinPeptideMatching)
{
kvp.Value.RemoveWhere(p => !parsimonyProteinList.ContainsKey(p.Protein));
}
return(ConstructProteinGroups(new HashSet <PeptideWithSetModifications>(proteinsWithUniquePeptides.Values.SelectMany(p => p)), new HashSet <PeptideWithSetModifications>(CompactPeptideToProteinPeptideMatching.Values.SelectMany(p => p))));
}