/// <summary>
/// Initializes digestion object
/// </summary>
/// <param name="digestionParams"></param>
/// <param name="allKnownFixedModifications"></param>
/// <param name="variableModifications"></param>
public ProteinDigestion(IDigestionParams digestionParams, IEnumerable <ModificationWithMass> allKnownFixedModifications, List <ModificationWithMass> variableModifications)
{
DigestionParams = digestionParams;
Protease = digestionParams.Protease;
MaximumMissedCleavages = digestionParams.MaxMissedCleavages;
InitiatorMethionineBehavior = digestionParams.InitiatorMethionineBehavior;
MinPeptidesLength = digestionParams.MinPeptideLength;
MaxPeptidesLength = digestionParams.MaxPeptideLength;
AllKnownFixedModifications = allKnownFixedModifications;
VariableModifications = variableModifications;
}
/// <summary>
/// Gets peptides for digestion of a protein
/// </summary>
/// <param name="protein"></param>
/// <param name="digestionParams"></param>
/// <param name="allKnownFixedModifications"></param>
/// <param name="variableModifications"></param>
/// <returns></returns>
public static IEnumerable <PeptideWithSetModifications> Digest(this Protein protein, IDigestionParams digestionParams, IEnumerable <ModificationWithMass> allKnownFixedModifications, List <ModificationWithMass> variableModifications)
{
ProteinDigestion digestion = new ProteinDigestion(digestionParams, allKnownFixedModifications, variableModifications);
return(digestionParams.SemiProteaseDigestion ? digestion.SemiSpecificDigestion(protein) : digestion.Digestion(protein));
}
/// <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 <ModificationWithMass> allKnownFixedModifications,
IDigestionParams digestionParams, List <ModificationWithMass> variableModifications)
{
int peptideLength = OneBasedEndResidueInProtein - OneBasedStartResidueInProtein + 1;
int maximumVariableModificationIsoforms = digestionParams.MaxModificationIsoforms;
int maxModsForPeptide = digestionParams.MaxModsForPeptide;
var twoBasedPossibleVariableAndLocalizeableModifications = new Dictionary <int, List <ModificationWithMass> >(peptideLength + 4);
var pepNTermVariableMods = new List <ModificationWithMass>();
twoBasedPossibleVariableAndLocalizeableModifications.Add(1, pepNTermVariableMods);
var pepCTermVariableMods = new List <ModificationWithMass>();
twoBasedPossibleVariableAndLocalizeableModifications.Add(peptideLength + 2, pepCTermVariableMods);
foreach (ModificationWithMass variableModification in variableModifications)
{
// Check if can be a n-term mod
if (CanBeNTerminalMod(variableModification, peptideLength))
{
pepNTermVariableMods.Add(variableModification);
}
for (int r = 0; r < peptideLength; r++)
{
if (Gptmd.GptmdEngine.ModFits(variableModification, Protein, r + 1, peptideLength, OneBasedStartResidueInProtein + r) &&
variableModification.terminusLocalization == TerminusLocalization.Any)
{
if (!twoBasedPossibleVariableAndLocalizeableModifications.TryGetValue(r + 2, out List <ModificationWithMass> residueVariableMods))
{
residueVariableMods = new List <ModificationWithMass> {
variableModification
};
twoBasedPossibleVariableAndLocalizeableModifications.Add(r + 2, residueVariableMods);
}
else
{
residueVariableMods.Add(variableModification);
}
}
}
// Check if can be a c-term mod
if (CanBeCTerminalMod(variableModification, peptideLength))
{
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 ModificationWithMass variableModification)
{
// Check if can be a n-term mod
if (locInPeptide == 1 && CanBeNTerminalMod(variableModification, peptideLength) && !Protein.IsDecoy)
{
pepNTermVariableMods.Add(variableModification);
}
for (int r = 0; r < peptideLength; r++)
{
if (locInPeptide == r + 1 &&
(Protein.IsDecoy ||
(Gptmd.GptmdEngine.ModFits(variableModification, Protein, r + 1, peptideLength, OneBasedStartResidueInProtein + r) &&
variableModification.terminusLocalization == TerminusLocalization.Any)))
{
if (!twoBasedPossibleVariableAndLocalizeableModifications.TryGetValue(r + 2, out List <ModificationWithMass> residueVariableMods))
{
residueVariableMods = new List <ModificationWithMass> {
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, ModificationWithMass> 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, OneBasedStartResidueInProtein, OneBasedEndResidueInProtein,
PeptideDescription, MissedCleavages, kvp, numFixedMods));
variable_modification_isoforms++;
if (variable_modification_isoforms == maximumVariableModificationIsoforms)
{
yield break;
}
}
}
