private SrmDocument AddProteins(SrmDocument document, ref IdentityPath selectedPath)
{
if (tabControl1.SelectedTab != tabPageProteinList)
return document;
var backgroundProteome = GetBackgroundProteome(document);
for (int i = gridViewProteins.Rows.Count - 1; i >= 0; i--)
{
var row = gridViewProteins.Rows[i];
var proteinName = Convert.ToString(row.Cells[colProteinName.Index].Value);
if (String.IsNullOrEmpty(proteinName))
{
continue;
}
var pastedMetadata = new ProteinMetadata(proteinName,
Convert.ToString(row.Cells[colProteinDescription.Index].Value),
NullForEmpty(Convert.ToString(row.Cells[colProteinPreferredName.Index].Value)),
NullForEmpty(Convert.ToString(row.Cells[colProteinAccession.Index].Value)),
NullForEmpty(Convert.ToString(row.Cells[colProteinGene.Index].Value)),
NullForEmpty(Convert.ToString(row.Cells[colProteinSpecies.Index].Value)));
FastaSequence fastaSequence = null;
if (!backgroundProteome.IsNone)
{
ProteinMetadata protdbMetadata;
fastaSequence = backgroundProteome.GetFastaSequence(proteinName, out protdbMetadata);
// Fill in any gaps in pasted metadata with that in protdb
pastedMetadata = pastedMetadata.Merge(protdbMetadata);
}
// Strip any whitespace (tab, newline etc) In case it was copied out of a FASTA file
var fastaSequenceString = new string(Convert.ToString(row.Cells[colProteinSequence.Index].Value).Where(c => !Char.IsWhiteSpace(c)).ToArray());
if (!string.IsNullOrEmpty(fastaSequenceString))
{
try
{
if (fastaSequence == null) // Didn't match anything in protdb
{
fastaSequence = new FastaSequence(pastedMetadata.Name, pastedMetadata.Description,
new ProteinMetadata[0], fastaSequenceString);
}
else
{
if (fastaSequence.Sequence != fastaSequenceString)
{
fastaSequence = new FastaSequence(pastedMetadata.Name, pastedMetadata.Description,
fastaSequence.Alternatives, fastaSequenceString);
}
}
}
catch (Exception exception)
{
ShowProteinError(new PasteError
{
Line = i,
Column = colProteinDescription.Index,
Message = string.Format(Resources.PasteDlg_AddProteins_Invalid_protein_sequence__0__, exception.Message)
});
return null;
}
}
if (fastaSequence == null)
{
ShowProteinError(
new PasteError
{
Line = i,
Message = backgroundProteome.IsNone
? Resources.PasteDlg_AddProteins_Missing_protein_sequence
: Resources.PasteDlg_AddProteins_This_protein_was_not_found_in_the_background_proteome_database
});
return null;
}
var description = pastedMetadata.Description;
if (!string.IsNullOrEmpty(description) && description != fastaSequence.Description)
{
fastaSequence = new FastaSequence(fastaSequence.Name, description, fastaSequence.Alternatives, fastaSequence.Sequence);
}
pastedMetadata = pastedMetadata.ChangeName(fastaSequence.Name).ChangeDescription(fastaSequence.Description); // Make sure these agree
var nodeGroupPep = new PeptideGroupDocNode(fastaSequence, pastedMetadata, new PeptideDocNode[0]);
nodeGroupPep = nodeGroupPep.ChangeSettings(document.Settings, SrmSettingsDiff.ALL);
var to = selectedPath;
if (to == null || to.Depth < (int)SrmDocument.Level.MoleculeGroups)
document = (SrmDocument)document.Add(nodeGroupPep);
else
{
Identity toId = selectedPath.GetIdentity((int)SrmDocument.Level.MoleculeGroups);
document = (SrmDocument)document.Insert(toId, nodeGroupPep);
}
selectedPath = new IdentityPath(nodeGroupPep.Id);
}
return document;
}
private static SrmDocument GenerateDecoysFunc(SrmDocument document, int numDecoys, bool multiCycle,
Func<SequenceMods, SequenceMods> genDecoySequence)
{
// Loop through the existing tree in random order creating decoys
var settings = document.Settings;
var enzyme = settings.PeptideSettings.Enzyme;
var decoyNodePepList = new List<PeptideDocNode>();
var setDecoyKeys = new HashSet<PeptideModKey>();
while (numDecoys > 0)
{
int startDecoys = numDecoys;
foreach (var nodePep in document.Peptides.ToArray().RandomOrder())
{
if (numDecoys == 0)
break;
// Decoys should not be based on standard peptides
if (nodePep.GlobalStandardType != null)
continue;
// If the non-terminal end of the peptide sequence is all a single character, skip this peptide,
// since it can't support decoy generation.
var sequence = nodePep.Peptide.Sequence;
if (genDecoySequence != null && sequence.Substring(0, sequence.Length - 1).Distinct().Count() == 1)
continue;
var seqMods = new SequenceMods(nodePep);
if (genDecoySequence != null)
{
seqMods = genDecoySequence(seqMods);
}
var peptide = nodePep.Peptide;
var decoyPeptide = new Peptide(null, seqMods.Sequence, null, null, enzyme.CountCleavagePoints(seqMods.Sequence), true);
if (seqMods.Mods != null)
seqMods.Mods = seqMods.Mods.ChangePeptide(decoyPeptide);
foreach (var comparableGroups in PeakFeatureEnumerator.ComparableGroups(nodePep))
{
var decoyNodeTranGroupList = GetDecoyGroups(nodePep, decoyPeptide, seqMods.Mods, comparableGroups, document,
Equals(seqMods.Sequence, peptide.Sequence));
if (decoyNodeTranGroupList.Count == 0)
continue;
var nodePepNew = new PeptideDocNode(decoyPeptide, settings, seqMods.Mods,
null, nodePep.ExplicitRetentionTime, decoyNodeTranGroupList.ToArray(), false);
if (!Equals(nodePep.ModifiedSequence, nodePepNew.ModifiedSequence))
{
var sourceKey = new ModifiedSequenceMods(nodePep.ModifiedSequence, nodePep.ExplicitMods);
nodePepNew = nodePepNew.ChangeSourceKey(sourceKey);
}
// Avoid adding duplicate peptides
if (setDecoyKeys.Contains(nodePepNew.Key))
continue;
setDecoyKeys.Add(nodePepNew.Key);
decoyNodePepList.Add(nodePepNew);
numDecoys--;
}
}
// Stop if not multi-cycle or the number of decoys has not changed.
if (!multiCycle || startDecoys == numDecoys)
break;
}
var decoyNodePepGroup = new PeptideGroupDocNode(new PeptideGroup(true), Annotations.EMPTY, PeptideGroup.DECOYS,
null, decoyNodePepList.ToArray(), false);
decoyNodePepGroup = decoyNodePepGroup.ChangeSettings(document.Settings, SrmSettingsDiff.ALL);
return (SrmDocument)document.Add(decoyNodePepGroup);
}