public PeptideGroupDocNode(PeptideGroup id, Annotations annotations, ProteinMetadata proteinMetadata,
PeptideDocNode[] children, bool autoManageChildren)
: base(id, annotations, children, autoManageChildren)
{
if (Equals(id.Name, proteinMetadata.Name))
proteinMetadata = proteinMetadata.ChangeName(null); // Make it clear that the name hasn't been altered
if (Equals(id.Description, proteinMetadata.Description))
proteinMetadata = proteinMetadata.ChangeDescription(null); // Make it clear that the description hasn't been altered
_proteinMetadata = proteinMetadata;
}
private PeptideGroupDocNode GetMoleculePeptideGroup(SrmDocument document, DataGridViewRow row, bool requireProductInfo)
{
var pepGroup = new PeptideGroup();
var pep = GetMoleculePeptide(document, row, pepGroup, requireProductInfo);
if (pep == null)
return null;
var name = Convert.ToString(row.Cells[INDEX_MOLECULE_GROUP].Value);
if (string.IsNullOrEmpty(name))
name = document.GetPeptideGroupId(true);
var metadata = new ProteinMetadata(name, string.Empty).SetWebSearchCompleted(); // FUTURE: some kind of lookup for small molecules
return new PeptideGroupDocNode(pepGroup, metadata, new[] {pep});
}
private PeptideDocNode GetMoleculePeptide(SrmDocument document, DataGridViewRow row, PeptideGroup group, bool requireProductInfo)
{
DocNodeCustomIon ion;
MoleculeInfo moleculeInfo;
try
{
moleculeInfo = ReadPrecursorOrProductColumns(document, row, true); // Re-read the precursor columns
if (moleculeInfo == null)
return null; // Some failure, but exception was already handled
ion = new DocNodeCustomIon(moleculeInfo.Formula, moleculeInfo.MonoMass, moleculeInfo.AverageMass,
Convert.ToString(row.Cells[INDEX_MOLECULE_NAME].Value)); // Short name
}
catch (ArgumentException e)
{
ShowTransitionError(new PasteError
{
Column = INDEX_MOLECULE_FORMULA,
Line = row.Index,
Message = e.Message
});
return null;
}
try
{
var pep = new Peptide(ion);
var tranGroup = GetMoleculeTransitionGroup(document, row, pep, requireProductInfo);
if (tranGroup == null)
return null;
return new PeptideDocNode(pep, document.Settings, null, null, moleculeInfo.ExplicitRetentionTime, new[] { tranGroup }, true);
}
catch (InvalidOperationException e)
{
ShowTransitionError(new PasteError
{
Column = INDEX_MOLECULE_FORMULA,
Line = row.Index,
Message = e.Message
});
return null;
}
}
public PeptideGroupDocNode ChangeSettings(SrmSettings settingsNew, SrmSettingsDiff diff)
{
if (diff.Monitor != null)
{
diff.Monitor.ProcessGroup(this);
}
if (diff.DiffPeptides && settingsNew.PeptideSettings.Filter.AutoSelect && AutoManageChildren)
{
IList <DocNode> childrenNew = new List <DocNode>();
int countPeptides = 0;
int countIons = 0;
Dictionary <int, DocNode> mapIndexToChild = CreateGlobalIndexToChildMap();
Dictionary <PeptideModKey, DocNode> mapIdToChild = CreatePeptideModToChildMap();
foreach (PeptideDocNode nodePep in GetPeptideNodes(settingsNew, true))
{
PeptideDocNode nodePepResult = nodePep;
SrmSettingsDiff diffNode = SrmSettingsDiff.ALL;
DocNode existing;
// Add values that existed before the change. First check for exact match by
// global index, which will happen when explicit modifications are added,
// and then by content identity.
if (mapIndexToChild.TryGetValue(nodePep.Id.GlobalIndex, out existing) ||
mapIdToChild.TryGetValue(nodePep.Key, out existing))
{
nodePepResult = (PeptideDocNode)existing;
diffNode = diff;
}
if (nodePepResult != null)
{
// Materialize children of the peptide.
nodePepResult = nodePepResult.ChangeSettings(settingsNew, diffNode);
childrenNew.Add(nodePepResult);
// Make sure a single peptide group does not exceed document limits.
countPeptides++;
countIons += nodePepResult.TransitionCount;
if (countIons > SrmDocument.MAX_TRANSITION_COUNT)
{
throw new InvalidDataException(String.Format(
Resources.PeptideGroupDocNode_ChangeSettings_The_current_document_settings_would_cause_the_number_of_targeted_transitions_to_exceed__0_n0___The_document_settings_must_be_more_restrictive_or_add_fewer_proteins_,
SrmDocument.MAX_TRANSITION_COUNT));
}
if (countPeptides > SrmDocument.MAX_PEPTIDE_COUNT)
{
throw new InvalidDataException(String.Format(
Resources.PeptideGroupDocNode_ChangeSettings_The_current_document_settings_would_cause_the_number_of_peptides_to_exceed__0_n0___The_document_settings_must_be_more_restrictive_or_add_fewer_proteins_,
SrmDocument.MAX_PEPTIDE_COUNT));
}
}
}
if (PeptideGroup.Sequence != null)
{
childrenNew = PeptideGroup.RankPeptides(childrenNew, settingsNew, true);
}
return((PeptideGroupDocNode)ChangeChildrenChecked(childrenNew));
}
else
{
var nodeResult = this;
if (diff.DiffPeptides && diff.SettingsOld != null)
{
// If variable modifications changed, remove all peptides with variable
// modifications which are no longer possible.
var modsNew = settingsNew.PeptideSettings.Modifications;
var modsVarNew = modsNew.VariableModifications.ToArray();
var modsOld = diff.SettingsOld.PeptideSettings.Modifications;
var modsVarOld = modsOld.VariableModifications.ToArray();
if (modsNew.MaxVariableMods < modsOld.MaxVariableMods ||
!ArrayUtil.EqualsDeep(modsVarNew, modsVarOld))
{
IList <DocNode> childrenNew = new List <DocNode>();
foreach (PeptideDocNode nodePeptide in nodeResult.Children)
{
if (nodePeptide.AreVariableModsPossible(modsNew.MaxVariableMods, modsVarNew))
{
childrenNew.Add(nodePeptide);
}
}
nodeResult = (PeptideGroupDocNode)nodeResult.ChangeChildrenChecked(childrenNew);
}
}
// Check for changes affecting children
if (diff.DiffPeptideProps || diff.DiffExplicit ||
diff.DiffTransitionGroups || diff.DiffTransitionGroupProps ||
diff.DiffTransitions || diff.DiffTransitionProps ||
diff.DiffResults)
{
IList <DocNode> childrenNew = new List <DocNode>();
// Enumerate the nodes making necessary changes.
foreach (PeptideDocNode nodePeptide in nodeResult.Children)
{
childrenNew.Add(nodePeptide.ChangeSettings(settingsNew, diff));
}
childrenNew = RankChildren(settingsNew, childrenNew);
nodeResult = (PeptideGroupDocNode)nodeResult.ChangeChildrenChecked(childrenNew);
}
return(nodeResult);
}
}
public SrmDocument ConvertToSmallMolecules(SrmDocument document,
ConvertToSmallMoleculesMode mode = ConvertToSmallMoleculesMode.formulas,
bool invertCharges = false,
bool ignoreDecoys=false)
{
if (mode == ConvertToSmallMoleculesMode.none)
return document;
var newdoc = new SrmDocument(document.Settings);
var note = new Annotations(TestingConvertedFromProteomic, null, 1); // Mark this as a testing node so we don't sort it
newdoc = (SrmDocument)newdoc.ChangeIgnoreChangingChildren(true); // Retain copied results
foreach (var peptideGroupDocNode in document.MoleculeGroups)
{
if (!peptideGroupDocNode.IsProteomic)
{
newdoc = (SrmDocument)newdoc.Add(peptideGroupDocNode); // Already a small molecule
}
else
{
var newPeptideGroup = new PeptideGroup();
var newPeptideGroupDocNode = new PeptideGroupDocNode(newPeptideGroup,
peptideGroupDocNode.Annotations.Merge(note), peptideGroupDocNode.Name,
peptideGroupDocNode.Description, new PeptideDocNode[0],
peptideGroupDocNode.AutoManageChildren);
foreach (var mol in peptideGroupDocNode.Molecules)
{
var peptideSequence = mol.Peptide.Sequence;
// Create a PeptideDocNode with the presumably baseline charge and label
var precursorCharge = (mol.TransitionGroups.Any() ? mol.TransitionGroups.First().TransitionGroup.PrecursorCharge : 0) * (invertCharges ? -1 : 1);
var isotopeLabelType = mol.TransitionGroups.Any() ? mol.TransitionGroups.First().TransitionGroup.LabelType : IsotopeLabelType.light;
var moleculeCustomIon = ConvertToSmallMolecule(mode, document, mol, precursorCharge, isotopeLabelType);
var precursorCustomIon = moleculeCustomIon;
var newPeptide = new Peptide(moleculeCustomIon);
var newPeptideDocNode = new PeptideDocNode(newPeptide, newdoc.Settings, null, null,
null, null, mol.ExplicitRetentionTime, note, mol.Results, new TransitionGroupDocNode[0],
mol.AutoManageChildren);
foreach (var transitionGroupDocNode in mol.TransitionGroups)
{
if (transitionGroupDocNode.IsDecoy)
{
if (ignoreDecoys)
continue;
throw new Exception("There is no translation from decoy to small molecules"); // Not L10N
}
if (transitionGroupDocNode.TransitionGroup.PrecursorCharge != Math.Abs(precursorCharge) ||
!Equals(isotopeLabelType, transitionGroupDocNode.TransitionGroup.LabelType))
{
// Different charges or labels mean different ion formulas
precursorCharge = transitionGroupDocNode.TransitionGroup.PrecursorCharge * (invertCharges ? -1 : 1);
isotopeLabelType = transitionGroupDocNode.TransitionGroup.LabelType;
precursorCustomIon = ConvertToSmallMolecule(mode, document, mol, precursorCharge, isotopeLabelType);
}
var newTransitionGroup = new TransitionGroup(newPeptide, precursorCustomIon, precursorCharge, isotopeLabelType);
// Remove any library info, since for the moment at least small molecules don't support this and it won't roundtrip
var resultsNew = RemoveTransitionGroupChromInfoLibraryInfo(transitionGroupDocNode);
var newTransitionGroupDocNode = new TransitionGroupDocNode(newTransitionGroup,
transitionGroupDocNode.Annotations.Merge(note), document.Settings,
null, null, transitionGroupDocNode.ExplicitValues, resultsNew, null,
transitionGroupDocNode.AutoManageChildren);
var mzShift = invertCharges ? 2.0 * BioMassCalc.MassProton : 0; // We removed hydrogen rather than added
Assume.IsTrue((Math.Abs(newTransitionGroupDocNode.PrecursorMz + mzShift - transitionGroupDocNode.PrecursorMz) - Math.Abs(transitionGroupDocNode.TransitionGroup.PrecursorCharge * BioMassCalc.MassElectron)) <= 1E-5);
foreach (var transition in transitionGroupDocNode.Transitions)
{
double mass = 0;
var transitionCharge = transition.Transition.Charge * (invertCharges ? -1 : 1);
var ionType = IonType.custom;
CustomIon transitionCustomIon;
double mzShiftTransition = 0;
if (transition.Transition.IonType == IonType.precursor)
{
ionType = IonType.precursor;
transitionCustomIon = new DocNodeCustomIon(precursorCustomIon.Formula,
string.IsNullOrEmpty(precursorCustomIon.Formula) ? precursorCustomIon.MonoisotopicMass : (double?) null,
string.IsNullOrEmpty(precursorCustomIon.Formula) ? precursorCustomIon.AverageMass : (double?) null,
SmallMoleculeNameFromPeptide(peptideSequence, transitionCharge));
mzShiftTransition = invertCharges ? 2.0 * BioMassCalc.MassProton : 0; // We removed hydrogen rather than added
}
else if (transition.Transition.IonType == IonType.custom)
{
transitionCustomIon = transition.Transition.CustomIon;
mass = transitionCustomIon.MonoisotopicMass;
}
else
{
// TODO - try to get fragment formula?
mass = BioMassCalc.CalculateIonMassFromMz(transition.Mz, transition.Transition.Charge);
transitionCustomIon = new DocNodeCustomIon(mass, mass,// We can't really get at mono vs average mass from m/z, but for test purposes this is fine
transition.Transition.FragmentIonName);
}
if (mode == ConvertToSmallMoleculesMode.masses_and_names)
{
// Discard the formula if we're testing the use of mass-with-names (for matching in ratio calcs) target specification
transitionCustomIon = new DocNodeCustomIon(transitionCustomIon.MonoisotopicMass, transitionCustomIon.AverageMass,
transition.Transition.FragmentIonName);
}
else if (mode == ConvertToSmallMoleculesMode.masses_only)
{
// Discard the formula and name if we're testing the use of mass-only target specification
transitionCustomIon = new DocNodeCustomIon(transitionCustomIon.MonoisotopicMass, transitionCustomIon.AverageMass);
}
var newTransition = new Transition(newTransitionGroup, ionType,
null, transition.Transition.MassIndex, transition.Transition.Charge * (invertCharges ? -1 : 1), null,
transitionCustomIon);
if (ionType == IonType.precursor)
{
mass = document.Settings.GetFragmentMass(transitionGroupDocNode.TransitionGroup.LabelType, null, newTransition, newTransitionGroupDocNode.IsotopeDist);
}
var newTransitionDocNode = new TransitionDocNode(newTransition, transition.Annotations.Merge(note),
null, mass, transition.IsotopeDistInfo, null,
transition.Results);
Assume.IsTrue((Math.Abs(newTransitionDocNode.Mz + mzShiftTransition - transition.Mz) - Math.Abs(transitionGroupDocNode.TransitionGroup.PrecursorCharge * BioMassCalc.MassElectron)) <= 1E-5, String.Format("unexpected mz difference {0}-{1}={2}", newTransitionDocNode.Mz , transition.Mz, newTransitionDocNode.Mz - transition.Mz)); // Not L10N
newTransitionGroupDocNode =
(TransitionGroupDocNode)newTransitionGroupDocNode.Add(newTransitionDocNode);
}
if (newPeptideDocNode != null)
newPeptideDocNode = (PeptideDocNode)newPeptideDocNode.Add(newTransitionGroupDocNode);
}
newPeptideGroupDocNode =
(PeptideGroupDocNode)newPeptideGroupDocNode.Add(newPeptideDocNode);
}
newdoc = (SrmDocument)newdoc.Add(newPeptideGroupDocNode);
}
}
// No retention time prediction for small molecules (yet?)
newdoc = newdoc.ChangeSettings(newdoc.Settings.ChangePeptideSettings(newdoc.Settings.PeptideSettings.ChangePrediction(
newdoc.Settings.PeptideSettings.Prediction.ChangeRetentionTime(null))));
return newdoc;
}
public PeptideGroupDocNode ChangeSettings(SrmSettings settingsNew, SrmSettingsDiff diff,
DocumentSettingsContext context = null)
{
if (diff.Monitor != null)
{
diff.Monitor.ProcessGroup(this);
}
if (diff.DiffPeptides && settingsNew.PeptideSettings.Filter.AutoSelect && AutoManageChildren)
{
IList <DocNode> childrenNew = new List <DocNode>();
int countPeptides = 0;
int countIons = 0;
Dictionary <int, DocNode> mapIndexToChild = CreateGlobalIndexToChildMap();
Dictionary <PeptideModKey, DocNode> mapIdToChild = CreatePeptideModToChildMap();
IEnumerable <PeptideDocNode> peptideDocNodes;
if (settingsNew.PeptideSettings.Filter.PeptideUniqueness == PeptideFilter.PeptideUniquenessConstraint.none ||
settingsNew.PeptideSettings.NeedsBackgroundProteomeUniquenessCheckProcessing)
{
peptideDocNodes = GetPeptideNodes(settingsNew, true).ToList();
}
else
{
// Checking peptide uniqueness against the background proteome can be expensive.
// Do all the regular processing, then filter those results at the end when we
// can do it in aggregate for best speed.
IEnumerable <PeptideDocNode> peptideDocNodesUnique;
var peptideDocNodesPrecalculatedForUniquenessCheck = context == null ? null : context.PeptideDocNodesPrecalculatedForUniquenessCheck;
var uniquenessDict = context == null ? null : context.UniquenessDict;
if (peptideDocNodesPrecalculatedForUniquenessCheck != null)
{
// Already processed, and a global list of peptides provided
Assume.IsNotNull(uniquenessDict);
peptideDocNodesUnique = peptideDocNodesPrecalculatedForUniquenessCheck;
}
else
{
// We'll have to do the processing for this node, and work with
// just the peptides on this node. With luck the background proteome
// will already have those cached for uniqueness checks.
var settingsNoUniquenessFilter =
settingsNew.ChangePeptideFilter(f => f.ChangePeptideUniqueness(PeptideFilter.PeptideUniquenessConstraint.none));
var nodes = GetPeptideNodes(settingsNoUniquenessFilter, true).ToList();
var sequences = new List <string>(from p in nodes select p.Peptide.Sequence);
peptideDocNodesUnique = nodes; // Avoid ReSharper multiple enumeration warning
uniquenessDict = settingsNew.PeptideSettings.Filter.CheckPeptideUniqueness(settingsNew, sequences, diff.Monitor);
}
// ReSharper disable once PossibleNullReferenceException
peptideDocNodes = peptideDocNodesUnique.Where(p =>
{
// It's possible during document load for uniqueness dict to get out of synch, so be
// cautious with lookup and just return false of not found. Final document change will clean that up.
bool isUnique;
return(uniquenessDict.TryGetValue(p.Peptide.Sequence, out isUnique) && isUnique);
});
}
foreach (var nodePep in peptideDocNodes)
{
PeptideDocNode nodePepResult = nodePep;
SrmSettingsDiff diffNode = SrmSettingsDiff.ALL;
DocNode existing;
// Add values that existed before the change. First check for exact match by
// global index, which will happen when explicit modifications are added,
// and then by content identity.
if (mapIndexToChild.TryGetValue(nodePep.Id.GlobalIndex, out existing) ||
mapIdToChild.TryGetValue(nodePep.Key, out existing))
{
nodePepResult = (PeptideDocNode)existing;
diffNode = diff;
}
if (nodePepResult != null)
{
// Materialize children of the peptide.
nodePepResult = nodePepResult.ChangeSettings(settingsNew, diffNode);
childrenNew.Add(nodePepResult);
// Make sure a single peptide group does not exceed document limits.
countPeptides++;
countIons += nodePepResult.TransitionCount;
if (countIons > SrmDocument.MAX_TRANSITION_COUNT)
{
throw new InvalidDataException(String.Format(
Resources.PeptideGroupDocNode_ChangeSettings_The_current_document_settings_would_cause_the_number_of_targeted_transitions_to_exceed__0_n0___The_document_settings_must_be_more_restrictive_or_add_fewer_proteins_,
SrmDocument.MAX_TRANSITION_COUNT));
}
if (countPeptides > SrmDocument.MAX_PEPTIDE_COUNT)
{
throw new InvalidDataException(String.Format(
Resources.PeptideGroupDocNode_ChangeSettings_The_current_document_settings_would_cause_the_number_of_peptides_to_exceed__0_n0___The_document_settings_must_be_more_restrictive_or_add_fewer_proteins_,
SrmDocument.MAX_PEPTIDE_COUNT));
}
}
}
if (PeptideGroup.Sequence != null)
{
childrenNew = PeptideGroup.RankPeptides(childrenNew, settingsNew, true);
}
return((PeptideGroupDocNode)ChangeChildrenChecked(childrenNew));
}
else
{
var nodeResult = this;
if (diff.DiffPeptides && diff.SettingsOld != null)
{
// If variable modifications changed, remove all peptides with variable
// modifications which are no longer possible.
var modsNew = settingsNew.PeptideSettings.Modifications;
var modsVarNew = modsNew.VariableModifications.ToArray();
var modsOld = diff.SettingsOld.PeptideSettings.Modifications;
var modsVarOld = modsOld.VariableModifications.ToArray();
if (modsNew.MaxVariableMods < modsOld.MaxVariableMods ||
!ArrayUtil.EqualsDeep(modsVarNew, modsVarOld))
{
IList <DocNode> childrenNew = new List <DocNode>();
foreach (PeptideDocNode nodePeptide in nodeResult.Children)
{
if (nodePeptide.AreVariableModsPossible(modsNew.MaxVariableMods, modsVarNew))
{
childrenNew.Add(nodePeptide);
}
}
nodeResult = (PeptideGroupDocNode)nodeResult.ChangeChildrenChecked(childrenNew);
}
}
// Check for changes affecting children
if (diff.DiffPeptideProps || diff.DiffExplicit ||
diff.DiffTransitionGroups || diff.DiffTransitionGroupProps ||
diff.DiffTransitions || diff.DiffTransitionProps ||
diff.DiffResults)
{
IList <DocNode> childrenNew = new List <DocNode>();
// Enumerate the nodes making necessary changes.
foreach (PeptideDocNode nodePeptide in nodeResult.Children)
{
childrenNew.Add(nodePeptide.ChangeSettings(settingsNew, diff));
}
childrenNew = RankChildren(settingsNew, childrenNew);
nodeResult = (PeptideGroupDocNode)nodeResult.ChangeChildrenChecked(childrenNew);
}
return(nodeResult);
}
}
public PeptideGroupDocNode(PeptideGroup id, Annotations annotations, string name, string description,
PeptideDocNode[] children, bool autoManageChildren)
: this(id, annotations, new ProteinMetadata(name, description), children, autoManageChildren)
{
}
public PeptideGroupDocNode(PeptideGroup id, Annotations annotations, string name, string description,
PeptideDocNode[] children)
: this(id, annotations, name, description, children, true)
{
}
public PeptideGroupDocNode(PeptideGroup id, ProteinMetadata proteinMetadata, PeptideDocNode[] children)
: this(id, Annotations.EMPTY, proteinMetadata, children, true)
{
}
private ProteinMetadata _proteinMetadata; // name, description, accession, gene, etc
#endregion Fields
#region Constructors
public PeptideGroupDocNode(PeptideGroup id, string name, string description, PeptideDocNode[] children)
: this(id, Annotations.EMPTY, name, description, children)
{
}
public PeptideGroupDocNode(PeptideGroup id, Annotations annotations, ProteinMetadata proteinMetadata,
PeptideDocNode[] children, bool autoManageChildren)
: this(id, annotations, proteinMetadata, children, autoManageChildren, null)
{
}