/// <summary>
/// Creates a PeptideWithSetModifications object from a sequence string.
/// Useful for reading in MetaMorpheus search engine output into mzLib objects.
/// </summary>
public PeptideWithSetModifications(string sequence, Dictionary <string, Modification> allKnownMods, int numFixedMods = 0,
DigestionParams digestionParams = null, Protein p = null, int oneBasedStartResidueInProtein = int.MinValue,
int oneBasedEndResidueInProtein = int.MinValue, int missedCleavages = int.MinValue,
CleavageSpecificity cleavageSpecificity = CleavageSpecificity.Full, string peptideDescription = null)
: base(p, oneBasedStartResidueInProtein, oneBasedEndResidueInProtein, missedCleavages, cleavageSpecificity, peptideDescription)
{
if (sequence.Contains("|"))
{
throw new MzLibUtil.MzLibException("Ambiguous peptide cannot be parsed from string: " + sequence);
}
FullSequence = sequence;
GetModsAfterDeserialization(allKnownMods, out _baseSequence);
NumFixedMods = numFixedMods;
_digestionParams = digestionParams;
if (p != null)
{
ProteinAccession = p.Accession;
}
if (digestionParams != null)
{
DigestionParamString = digestionParams.ToString();
}
}
/// <summary>
/// Initializes digestion object
/// </summary>
/// <param name="digestionParams"></param>
/// <param name="allKnownFixedModifications"></param>
/// <param name="variableModifications"></param>
public ProteinDigestion(DigestionParams digestionParams, IEnumerable <Modification> allKnownFixedModifications, List <Modification> variableModifications)
{
DigestionParams = digestionParams;
Protease = digestionParams.Protease;
MaximumMissedCleavages = digestionParams.MaxMissedCleavages;
InitiatorMethionineBehavior = digestionParams.InitiatorMethionineBehavior;
MinPeptidesLength = digestionParams.MinPeptideLength;
MaxPeptidesLength = digestionParams.MaxPeptideLength;
AllKnownFixedModifications = allKnownFixedModifications ?? new List <Modification>();
VariableModifications = variableModifications ?? new List <Modification>();
}
private readonly string ProteinAccession; // used to get protein object after deserialization
/// <summary>
/// Creates a PeptideWithSetModifications object from a protein. Used when a Protein is digested.
/// </summary>
public PeptideWithSetModifications(Protein protein, DigestionParams digestionParams, int oneBasedStartResidueInProtein,
int oneBasedEndResidueInProtein, CleavageSpecificity cleavageSpecificity, string peptideDescription, int missedCleavages,
Dictionary <int, Modification> allModsOneIsNterminus, int numFixedMods, string baseSequence = null)
: base(protein, oneBasedStartResidueInProtein, oneBasedEndResidueInProtein, missedCleavages, cleavageSpecificity, peptideDescription, baseSequence)
{
_allModsOneIsNterminus = allModsOneIsNterminus;
NumFixedMods = numFixedMods;
_digestionParams = digestionParams;
DetermineFullSequence();
ProteinAccession = protein.Accession;
UpdateCleavageSpecificity();
}
private readonly string ProteinAccession; // used to get protein object after deserialization
/// <summary>
/// Creates a PeptideWithSetModifications object from a protein. Used when a Protein is digested.
/// </summary>
public PeptideWithSetModifications(Protein protein, DigestionParams digestionParams, int oneBasedStartResidueInProtein,
int oneBasedEndResidueInProtein, string peptideDescription, int missedCleavages,
Dictionary <int, Modification> allModsOneIsNterminus, int numFixedMods)
: base(protein, oneBasedStartResidueInProtein, oneBasedEndResidueInProtein, missedCleavages, peptideDescription)
{
_allModsOneIsNterminus = allModsOneIsNterminus;
NumFixedMods = numFixedMods;
_digestionParams = digestionParams;
DetermineFullSequence();
ProteinAccession = protein.Accession;
DigestionParamString = digestionParams.ToString();
}
public override bool Equals(object obj)
{
DigestionParams a = obj as DigestionParams;
return(a != null &&
this.MaxMissedCleavages.Equals(a.MaxMissedCleavages) &&
this.MinPeptideLength.Equals(a.MinPeptideLength) &&
this.MaxPeptideLength.Equals(a.MaxPeptideLength) &&
this.InitiatorMethionineBehavior.Equals(a.InitiatorMethionineBehavior) &&
this.MaxModificationIsoforms.Equals(a.MaxModificationIsoforms) &&
this.MaxModsForPeptide.Equals(a.MaxModsForPeptide) &&
this.Protease.Equals(a.Protease) &&
this.SemiProteaseDigestion.Equals(a.SemiProteaseDigestion) &&
this.TerminusTypeSemiProtease.Equals(a.TerminusTypeSemiProtease));
}
/// <summary>
/// This should be run after deserialization of a PeptideWithSetModifications, in order to set its Protein and Modification objects, which were not serialized
/// </summary>
public void SetNonSerializedPeptideInfo(Dictionary <string, Modification> idToMod, Dictionary <string, Protein> accessionToProtein, DigestionParams dp)
{
GetModsAfterDeserialization(idToMod);
GetProteinAfterDeserialization(accessionToProtein);
_digestionParams = dp;
}
/// <summary>
/// Gets the peptides for a specific protein interval
/// </summary>
/// <param name="interval"></param>
/// <param name="allKnownFixedModifications"></param>
/// <param name="digestionParams"></param>
/// <param name="variableModifications"></param>
/// <returns></returns>
internal IEnumerable <PeptideWithSetModifications> GetModifiedPeptides(IEnumerable <Modification> allKnownFixedModifications,
DigestionParams digestionParams, List <Modification> variableModifications)
{
int peptideLength = OneBasedEndResidueInProtein - OneBasedStartResidueInProtein + 1;
int maximumVariableModificationIsoforms = digestionParams.MaxModificationIsoforms;
int maxModsForPeptide = digestionParams.MaxModsForPeptide;
var twoBasedPossibleVariableAndLocalizeableModifications = new Dictionary <int, List <Modification> >(peptideLength + 4);
var pepNTermVariableMods = new List <Modification>();
twoBasedPossibleVariableAndLocalizeableModifications.Add(1, pepNTermVariableMods);
var pepCTermVariableMods = new List <Modification>();
twoBasedPossibleVariableAndLocalizeableModifications.Add(peptideLength + 2, pepCTermVariableMods);
foreach (Modification variableModification in variableModifications)
{
// Check if can be a n-term mod
if (CanBeNTerminalMod(variableModification, peptideLength) && !ModificationLocalization.UniprotModExists(Protein, 1, variableModification))
{
pepNTermVariableMods.Add(variableModification);
}
for (int r = 0; r < peptideLength; r++)
{
if (ModificationLocalization.ModFits(variableModification, Protein.BaseSequence, r + 1, peptideLength, OneBasedStartResidueInProtein + r) &&
variableModification.LocationRestriction == "Anywhere." && !ModificationLocalization.UniprotModExists(Protein, r + 1, variableModification))
{
if (!twoBasedPossibleVariableAndLocalizeableModifications.TryGetValue(r + 2, out List <Modification> residueVariableMods))
{
residueVariableMods = new List <Modification> {
variableModification
};
twoBasedPossibleVariableAndLocalizeableModifications.Add(r + 2, residueVariableMods);
}
else
{
residueVariableMods.Add(variableModification);
}
}
}
// Check if can be a c-term mod
if (CanBeCTerminalMod(variableModification, peptideLength) && !ModificationLocalization.UniprotModExists(Protein, peptideLength, variableModification))
{
pepCTermVariableMods.Add(variableModification);
}
}
// LOCALIZED MODS
foreach (var kvp in Protein.OneBasedPossibleLocalizedModifications)
{
bool inBounds = kvp.Key >= OneBasedStartResidueInProtein && kvp.Key <= OneBasedEndResidueInProtein;
if (!inBounds)
{
continue;
}
int locInPeptide = kvp.Key - OneBasedStartResidueInProtein + 1;
foreach (Modification modWithMass in kvp.Value)
{
if (modWithMass is Modification variableModification)
{
// Check if can be a n-term mod
if (locInPeptide == 1 && CanBeNTerminalMod(variableModification, peptideLength) && !Protein.IsDecoy)
{
pepNTermVariableMods.Add(variableModification);
}
int r = locInPeptide - 1;
if (r >= 0 && r < peptideLength &&
(Protein.IsDecoy ||
(ModificationLocalization.ModFits(variableModification, Protein.BaseSequence, r + 1, peptideLength, OneBasedStartResidueInProtein + r) &&
variableModification.LocationRestriction == "Anywhere.")))
{
if (!twoBasedPossibleVariableAndLocalizeableModifications.TryGetValue(r + 2, out List <Modification> residueVariableMods))
{
residueVariableMods = new List <Modification> {
variableModification
};
twoBasedPossibleVariableAndLocalizeableModifications.Add(r + 2, residueVariableMods);
}
else
{
residueVariableMods.Add(variableModification);
}
}
// Check if can be a c-term mod
if (locInPeptide == peptideLength && CanBeCTerminalMod(variableModification, peptideLength) && !Protein.IsDecoy)
{
pepCTermVariableMods.Add(variableModification);
}
}
}
}
int variable_modification_isoforms = 0;
foreach (Dictionary <int, Modification> kvp in GetVariableModificationPatterns(twoBasedPossibleVariableAndLocalizeableModifications, maxModsForPeptide, peptideLength))
{
int numFixedMods = 0;
foreach (var ok in GetFixedModsOneIsNterminus(peptideLength, allKnownFixedModifications))
{
if (!kvp.ContainsKey(ok.Key))
{
numFixedMods++;
kvp.Add(ok.Key, ok.Value);
}
}
yield return(new PeptideWithSetModifications(Protein, digestionParams, OneBasedStartResidueInProtein, OneBasedEndResidueInProtein,
CleavageSpecificityForFdrCategory, PeptideDescription, MissedCleavages, kvp, numFixedMods));
variable_modification_isoforms++;
if (variable_modification_isoforms == maximumVariableModificationIsoforms)
{
yield break;
}
}
}