public void TestParsingManyMods()
{
const string modFilePath = @"\\protoapps\UserData\Jungkap\Lewy\db\Mods.txt";
var aaSet = new AminoAcidSet(modFilePath);
//aaSet.Display();
//SequenceLocation.ProteinNTerm
var residue = AminoAcid.ProteinNTerm.Residue;
var location = SequenceLocation.ProteinNTerm;
var aa = aaSet.GetAminoAcid(residue, location);
Console.Write("{0}\t{1}\t{2}", residue, aa.Mass, aa.Composition);
foreach (var modIndex in aaSet.GetModificationIndices(residue, location))
{
var modification = aaSet.GetModificationParams().GetModification(modIndex);
Console.WriteLine(modification.Mass);
//Console.Write("\t" + _modificationParams.GetModification(modIndex));
}
Console.WriteLine();
residue = AminoAcid.ProteinCTerm.Residue;
location = SequenceLocation.ProteinCTerm;
aa = aaSet.GetAminoAcid(residue, location);
Console.Write("{0}\t{1}\t{2}", residue, aa.Mass, aa.Composition);
foreach (var modIndex in aaSet.GetModificationIndices(residue, location))
{
var modification = aaSet.GetModificationParams().GetModification(modIndex);
Console.WriteLine(modification.Mass);
//Console.Write("\t" + _modificationParams.GetModification(modIndex));
}
//foreach (var aa in AminoAcid.StandardAminoAcidArr)
/*
var keys = _locationSpecificResidueMap[location].Keys.ToArray();
Array.Sort(keys);
foreach (var residue in keys)
{
var aa = GetAminoAcid(residue, location);
Console.Write("{0}\t{1}\t{2}", residue, aa.Mass, aa.Composition);
foreach (var modIndex in GetModificationIndices(residue, location))
{
Console.Write("\t" + _modificationParams.GetModification(modIndex));
}
Console.WriteLine();
}
} */
}
/// <summary>
/// Create a sequence using the supplied character sequence, modifications, and amino acid set
/// </summary>
/// <param name="sequence"></param>
/// <param name="modStr">E.g. Acetyl 0,Oxidation 1,Oxidation 20,Oxidation 27</param>
/// <param name="aminoAcidSet"></param>
/// <returns></returns>
public static Sequence CreateSequence(string sequence, string modStr, AminoAcidSet aminoAcidSet)
{
if (string.IsNullOrEmpty(modStr))
{
return(new Sequence(sequence, aminoAcidSet));
}
var indexModMap = new Dictionary <int, Modification>();
foreach (var modIns in modStr.Split(','))
{
var token = modIns.Split(' ');
if (token.Length != 2)
{
return(null); // invalid modStr
}
var mod = Modification.Get(token[0]);
if (mod == null)
{
return(null);
}
var index = Convert.ToInt32(token[1]) - 1;
indexModMap.Add(index, mod);
}
var aaList = new List <AminoAcid>();
for (var i = 0; i < sequence.Length; i++)
{
var residue = sequence[i];
var aa = aminoAcidSet.GetAminoAcid(residue);
if (i == 0 && indexModMap.ContainsKey(-1)) // N-term modification
{
var nTermMod = indexModMap[-1];
aa = new ModifiedAminoAcid(aa, nTermMod);
}
if (indexModMap.TryGetValue(i, out var mod))
{
var modifiedAa = new ModifiedAminoAcid(aa, mod);
aaList.Add(modifiedAa);
}
else
{
aaList.Add(aa);
}
}
return(new Sequence(aaList));
}
/// <summary>
/// Add an amino acid residue to this generator.
/// </summary>
/// <param name="index">index to add the amino acid. 0 is C-term. 1 is the C-term amino acid.</param>
/// <param name="residue">amino acid residue to add.</param>
/// <returns>true if residue is a valid amino acid; false otherwise.</returns>
private bool PutAminoAcid(int index, char residue)
{
_index = index + 1;
SequenceLocation?location = null;
if (_index == 1) // C-term residue
{
if (residue == AminoAcid.PeptideCTerm.Residue)
{
location = SequenceLocation.PeptideCTerm;
}
else if (residue == AminoAcid.ProteinCTerm.Residue)
{
location = SequenceLocation.ProteinCTerm;
}
}
else if (_index == _aminoAcidSequence.Length - 1 - NumNTermCleavages) // N-term residue
{
if (residue == AminoAcid.PeptideNTerm.Residue)
{
location = SequenceLocation.PeptideNTerm;
}
else if (residue == AminoAcid.ProteinNTerm.Residue)
{
location = SequenceLocation.ProteinNTerm;
}
}
else if (_index == 2) // Amino acid at the C-term
{
if (_aminoAcidSequence[1] == AminoAcid.PeptideCTerm)
{
location = SequenceLocation.PeptideCTerm;
}
else if (_aminoAcidSequence[1] == AminoAcid.ProteinCTerm)
{
location = SequenceLocation.ProteinCTerm;
}
}
else if (_index == _aminoAcidSequence.Length - 2 - NumNTermCleavages) // Amino acid at the N-term
{
if (_aminoAcidSequence[_aminoAcidSequence.Length - 1] == AminoAcid.PeptideNTerm)
{
location = SequenceLocation.PeptideNTerm;
}
else if (_aminoAcidSequence[_aminoAcidSequence.Length - 1] == AminoAcid.ProteinNTerm)
{
location = SequenceLocation.ProteinNTerm;
}
}
else
{
location = SequenceLocation.Everywhere;
}
if (location == null)
{
return(false);
}
var loc = (SequenceLocation)location;
var aminoAcid = AminoAcidSet.GetAminoAcid(residue, loc);
if (aminoAcid == null) // residue is not valid
{
return(false);
}
_aminoAcidSequence[_index] = aminoAcid;
_suffixComposition[_index] = _suffixComposition[_index - 1] + aminoAcid.Composition;
var modIndices = AminoAcidSet.GetModificationIndices(residue, loc);
if (!modIndices.Any()) // No modification
{
_graph[_index] = new Node[_graph[_index - 1].Length];
for (var i = 0; i < _graph[_index - 1].Length; i++)
{
_graph[_index][i] = new Node(_graph[_index - 1][i].ModificationCombinationIndex, i);
}
}
else
{
var modCombIndexToNodeMap = new Dictionary <int, Node>();
for (var i = 0; i < _graph[_index - 1].Length; i++)
{
var prevNodeIndex = i;
var prevNode = _graph[_index - 1][i];
var prevModCombIndex = prevNode.ModificationCombinationIndex;
// unmodified edge
if (modCombIndexToNodeMap.TryGetValue(prevModCombIndex, out var unmodifiedEdgeNode))
{
unmodifiedEdgeNode.AddPrevNodeIndex(prevNodeIndex);
}
else
{
modCombIndexToNodeMap.Add(prevModCombIndex, new Node(prevModCombIndex, prevNodeIndex));
}
// modified edge
foreach (var modIndex in modIndices)
{
var modCombIndex = ModificationParams.GetModificationCombinationIndex(
prevNode.ModificationCombinationIndex, modIndex);
if (modCombIndex < 0) // too many modifications
{
continue;
}
if (modCombIndexToNodeMap.TryGetValue(modCombIndex, out var modifiedEdgeNode))
{
modifiedEdgeNode.AddPrevNodeIndex(prevNodeIndex);
}
else
{
modCombIndexToNodeMap.Add(modCombIndex, new Node(modCombIndex, prevNodeIndex));
}
}
_graph[_index] = modCombIndexToNodeMap.Values.ToArray();
}
}
return(true);
}
static MgfSequenceReader()
{
StandardAminoAcidSet = new AminoAcidSet(Modification.Carbamidomethylation);
Modifications = new Dictionary<string, Tuple<AminoAcid, List<Modification>>>();
Modifications.Add("99.032",
new Tuple<AminoAcid, List<Modification>>(StandardAminoAcidSet.GetAminoAcid('G'),
new List<Modification> { Modification.Acetylation }));
Modifications.Add("113.048",
new Tuple<AminoAcid, List<Modification>>(StandardAminoAcidSet.GetAminoAcid('A'),
new List<Modification> { Modification.Acetylation }));
Modifications.Add("129.043",
new Tuple<AminoAcid, List<Modification>>(StandardAminoAcidSet.GetAminoAcid('S'),
new List<Modification> { Modification.Acetylation }));
Modifications.Add("141.079",
new Tuple<AminoAcid, List<Modification>>(StandardAminoAcidSet.GetAminoAcid('V'),
new List<Modification> { Modification.Acetylation }));
Modifications.Add("143.059",
new Tuple<AminoAcid, List<Modification>>(StandardAminoAcidSet.GetAminoAcid('T'),
new List<Modification> { Modification.Acetylation }));
Modifications.Add("147.035",
new Tuple<AminoAcid, List<Modification>>(StandardAminoAcidSet.GetAminoAcid('M'),
new List<Modification> { Modification.Oxidation }));
Modifications.Add("157.038",
new Tuple<AminoAcid, List<Modification>>(StandardAminoAcidSet.GetAminoAcid('D'),
new List<Modification> { Modification.Acetylation }));
Modifications.Add("160.03",
new Tuple<AminoAcid, List<Modification>>(StandardAminoAcidSet.GetAminoAcid('C'),
new List<Modification> { Modification.Carbamidomethylation }));
Modifications.Add("171.054",
new Tuple<AminoAcid, List<Modification>>(StandardAminoAcidSet.GetAminoAcid('E'),
new List<Modification> { Modification.Acetylation }));
Modifications.Add("173.051",
new Tuple<AminoAcid, List<Modification>>(StandardAminoAcidSet.GetAminoAcid('M'),
new List<Modification> { Modification.Acetylation }));
Modifications.Add("189.046",
new Tuple<AminoAcid, List<Modification>>(StandardAminoAcidSet.GetAminoAcid('F'),
new List<Modification> { Modification.Acetylation }));
Modifications.Add("202.041",
new Tuple<AminoAcid, List<Modification>>(StandardAminoAcidSet.GetAminoAcid('C'),
new List<Modification> { Modification.Carbamidomethylation,
Modification.Acetylation }));
}
/// <summary>
/// Add an amino acid residue to this generator.
/// </summary>
/// <param name="index">index to add the amino acid. 0 is C-term. 1 is the C-term amino acid.</param>
/// <param name="residue">amino acid residue to add.</param>
/// <param name="loc">location of the residue</param>
/// <returns>true if residue is a valid amino acid; false otherwise.</returns>
private bool PutAminoAcid(int index, char residue, SequenceLocation loc)
{
_index = index + 1;
var aminoAcid = AminoAcidSet.GetAminoAcid(residue, loc);
if (aminoAcid == null) // residue is not valid
{
return(false);
}
var fragmentComposition = _fragmentComposition[_index - 1] + aminoAcid.Composition;
if (fragmentComposition.Mass > _maxSequenceMass)
{
return(false);
}
_aminoAcidSequence[_index] = aminoAcid;
_fragmentComposition[_index] = fragmentComposition;
var modIndices = AminoAcidSet.GetModificationIndices(residue, loc);
if (!modIndices.Any()) // No modification
{
_graph[_index] = new Node[_graph[_index - 1].Length];
for (var i = 0; i < _graph[_index - 1].Length; i++)
{
_graph[_index][i] = new Node(_graph[_index - 1][i].ModificationCombinationIndex, i);
}
}
else
{
var modCombIndexToNodeMap = new Dictionary <int, Node>();
for (var i = 0; i < _graph[_index - 1].Length; i++)
{
var prevNodeIndex = i;
var prevNode = _graph[_index - 1][i];
var prevModCombIndex = prevNode.ModificationCombinationIndex;
// unmodified edge
if (modCombIndexToNodeMap.TryGetValue(prevModCombIndex, out var unmodifiedEdgeNode))
{
unmodifiedEdgeNode.AddPrevNodeIndex(prevNodeIndex);
}
else
{
modCombIndexToNodeMap.Add(prevModCombIndex, new Node(prevModCombIndex, prevNodeIndex));
}
// modified edge
foreach (var modIndex in modIndices)
{
var modCombIndex = ModificationParams.GetModificationCombinationIndex(
prevNode.ModificationCombinationIndex, modIndex);
if (modCombIndex < 0) // too many modifications
{
continue;
}
if (modCombIndexToNodeMap.TryGetValue(modCombIndex, out var modifiedEdgeNode))
{
modifiedEdgeNode.AddPrevNodeIndex(prevNodeIndex);
}
else
{
modCombIndexToNodeMap.Add(modCombIndex, new Node(modCombIndex, prevNodeIndex));
}
}
_graph[_index] = modCombIndexToNodeMap.Values.ToArray();
}
}
return(true);
}
private int GetNominalMass(AminoAcidSet aaSet, string peptide)
{
var nominalMass = 0;
StringBuilder buf = null;
var curNominalMass = 0;
foreach (var c in peptide)
{
if (char.IsUpper(c))
{
curNominalMass = aaSet.GetAminoAcid(c).GetNominalMass();
nominalMass += curNominalMass;
}
else if (c == '[') buf = new StringBuilder();
else if (char.IsNumber(c) && buf != null) buf.Append(c);
else if (c == ']' && buf != null) nominalMass += Convert.ToInt32(buf.ToString()) - curNominalMass;
}
return nominalMass;
}
