/// <summary>
/// Gets the sequence.
/// </summary>
/// <param name="proteoform">The proteoform.</param>
/// <returns></returns>
public static string?GetSequence(this IProteoformGroup proteoform)
{
if (proteoform?.Residues == null)
{
return(null);
}
return(new string(proteoform.Residues.Select(x => x.Symbol).ToArray()));
}
/// <summary>Generates a chemical proteoform hash for the specified proteoform.</summary>
/// <param name="proForma">The ProForma.</param>
/// <returns></returns>
public IChemicalProteoformHash Generate(string proForma)
{
// Parse string into term
ProFormaTerm originalProFormaTerm = this._proFormaParser.ParseString(proForma);
// Check to see if this is only a sequence
if (originalProFormaTerm.NTerminalDescriptors == null &&
originalProFormaTerm.CTerminalDescriptors == null &&
originalProFormaTerm.Tags == null &&
originalProFormaTerm.GlobalModifications == null &&
originalProFormaTerm.LabileDescriptors == null &&
originalProFormaTerm.TagGroups == null &&
originalProFormaTerm.UnlocalizedTags == null)
{
IProteoformGroup simpleGroup = this._proteoformGroupFactory.CreateProteoformGroup(originalProFormaTerm.Sequence);
return(new ChemicalProteoformHash(originalProFormaTerm.Sequence, simpleGroup));
}
// Create proteoform group (flattens all features into Ids)
IProteoformGroup proteoformGroup = this._proteoformGroupFactory.CreateProteoformGroup(originalProFormaTerm,
this._proteoformModificationLookup);
ProFormaDescriptor?nTermDescriptor =
this.CreateDescriptor(proteoformGroup.NTerminalModification);
IList <ProFormaDescriptor>?nTermDescriptors = nTermDescriptor == null ? null : new[] { nTermDescriptor };
ProFormaDescriptor?cTermDescriptor =
this.CreateDescriptor(proteoformGroup.CTerminalModification);
IList <ProFormaDescriptor>?cTermDescriptors = cTermDescriptor == null ? null : new[] { cTermDescriptor };
IList <ProFormaTag>? tags = null;
IList <ProFormaDescriptor>? labileDescriptors = null;
IList <ProFormaUnlocalizedTag>? unlocalizedTags = null;
IList <ProFormaTagGroup>? tagGroups = null;
IList <ProFormaGlobalModification>?globalModifications = null;
if (proteoformGroup.LocalizedModifications?.Count > 0)
{
foreach (var mod in proteoformGroup.LocalizedModifications)
{
ProFormaDescriptor?descriptor = this.CreateDescriptor(mod.ModificationDelta);
if (descriptor != null)
{
tags ??= new List <ProFormaTag>();
tags.Add(new ProFormaTag(mod.ZeroBasedStartIndex, mod.ZeroBasedEndIndex, new[] { descriptor }));
}
}
}
if (proteoformGroup.UnlocalizedModifications?.Count > 0)
{
foreach (var mod in proteoformGroup.UnlocalizedModifications)
{
ProFormaDescriptor?descriptor = this.CreateDescriptor(mod.ModificationDelta);
if (descriptor != null)
{
if (mod.IsLabile)
{
labileDescriptors ??= new List <ProFormaDescriptor>();
for (int i = 0; i < mod.Count; i++)
{
labileDescriptors.Add(descriptor);
}
}
else
{
unlocalizedTags ??= new List <ProFormaUnlocalizedTag>();
unlocalizedTags.Add(new ProFormaUnlocalizedTag(mod.Count, new[] { descriptor }));
}
}
}
}
if (proteoformGroup.ModificationGroups?.Count > 0)
{
foreach (var mod in proteoformGroup.ModificationGroups)
{
ProFormaDescriptor?descriptor = this.CreateDescriptor(mod.ModificationDelta);
if (descriptor != null)
{
tagGroups ??= new List <ProFormaTagGroup>();
tagGroups.Add(new ProFormaTagGroup(mod.GroupName, descriptor.Key, descriptor.EvidenceType, descriptor.Value,
mod.Members.Select(x => new ProFormaMembershipDescriptor(x.ZeroBasedStartIndex, x.ZeroBasedEndIndex, x.Weight)).ToList()));
}
}
}
if (proteoformGroup.GlobalModifications?.Count > 0)
{
foreach (var mod in proteoformGroup.GlobalModifications)
{
ProFormaDescriptor?descriptor = this.CreateDescriptor(mod.ModificationDelta);
if (descriptor != null)
{
globalModifications ??= new List <ProFormaGlobalModification>();
globalModifications.Add(new ProFormaGlobalModification(new[] { descriptor }, mod.TargetAminoAcids));
}
}
}
string?sequence = proteoformGroup.GetSequence();
if (sequence != null)
{
ProFormaTerm proFormaTerm = new(sequence, tags : tags?.OrderBy(t => t.Descriptors.First().Value).ToArray(),
nTerminalDescriptors : nTermDescriptors,
cTerminalDescriptors : cTermDescriptors,
labileDescriptors : labileDescriptors,
unlocalizedTags : unlocalizedTags,
tagGroups : tagGroups,
globalModifications : globalModifications);
string hash = new ProFormaWriter().WriteString(proFormaTerm);
return(new ChemicalProteoformHash(hash, proteoformGroup));
}
throw new Exception("Cannot get amino acid sequence for the proteoform group.");
}