private static void AddModification(string modMass, char residue, Modification modType)
{
var aminoAcid = StandardAminoAcidSet.GetAminoAcid(residue);
var modList = new Tuple <AminoAcid, List <Modification> >(aminoAcid, new List <Modification> {
modType
});
Modifications.Add(modMass, modList);
}
public void TestParsingManyMods(string modDefsFile)
{
var methodName = MethodBase.GetCurrentMethod().Name;
var modFile = Utils.GetTestFile(methodName, modDefsFile.Replace("TEST_FOLDER", Utils.DEFAULT_TEST_FILE_FOLDER));
if (!modFile.Exists)
{
Assert.Ignore("Ignoring test TestParsingManyMods since file not found: " + modFile.FullName);
}
var aaSet = new AminoAcidSet(modFile.FullName);
//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();
* }*/
//}
}
static BaseIonType()
{
// For A, B, C, X, Y, Z ions: sum of all amino acids + offset
A = new BaseIonType("a", true, new Composition(-1, 0, 0, -1, 0)); // -CO
Ar = new BaseIonType("a.", true, A.OffsetComposition + Composition.Hydrogen);
B = new BaseIonType("b", true, Composition.Zero);
C = new BaseIonType("c", true, Composition.NH3);
//X = new BaseIonType("x", false, Composition.H2O + Composition.CO);
X = new BaseIonType("x", false, new CompositionWithDeltaMass(44.9977) - Composition.Hydrogen);
Xr = new BaseIonType("x.", false, X.OffsetComposition + Composition.Hydrogen);
Y = new BaseIonType("y", false, Composition.H2O);
YM1 = new BaseIonType("y-1", false, Y.OffsetComposition - Composition.Hydrogen);
Z = new BaseIonType("z", false, Composition.H2O - Composition.NH2);
Zr = new BaseIonType("z.", false, Z.OffsetComposition + Composition.Hydrogen);
var aminoAcidSet = new AminoAcidSet();
// D ions have additional options for isoleucine and threonine. All offsets defined as sum of previous residues + offset.
var dIonOffsets = new Dictionary <char, double[]>
{ // Defined in terms of sum of previous residue weights + offset
{ '*', new [] { 44.0500 - Composition.Hydrogen.Mass } }, // For all residues except for V, I and T
{ 'V', new [] { 58.0657 - Composition.Hydrogen.Mass } },
{ 'I', new [] { 58.0657 - Composition.Hydrogen.Mass, 72.0813 - Composition.Hydrogen.Mass } }, // for isoleucine
{ 'T', new [] { 58.0657 - Composition.Hydrogen.Mass, 60.0450 - Composition.Hydrogen.Mass } } // for threonine
};
D = new BaseIonType("d", true, new CompositionWithDeltaMass(44.0500), aminoAcid => DefaultCompositionCalculator(dIonOffsets, aminoAcid));
// V only has one option for all terminal residues. Sum of previous residues + offset
V = new BaseIonType("v", false, new CompositionWithDeltaMass(74.0242), r => new List <Composition> {
r == null ? new CompositionWithDeltaMass(74.0242) : new CompositionWithDeltaMass(74.0242) - r.Composition - Composition.Hydrogen
});
// W ions have additional options for isoleucine and threonine. All offsets defined as sum of previous residues + offset.
var wIonOffsets = new Dictionary <char, double[]>
{ // Defined in terms of sum of previous residue weights + offset
{ '*', new [] { 73.0290 - Composition.Hydrogen.Mass } }, // For all residues except for V, I and T
{ 'V', new [] { 87.0446 - Composition.Hydrogen.Mass } },
{ 'I', new [] { 87.0446 - Composition.Hydrogen.Mass, aminoAcidSet.GetAminoAcid('I').Mass - 12.0238 - Composition.Hydrogen.Mass } }, // for isoleucine
{ 'T', new [] { 87.0446 - Composition.Hydrogen.Mass, aminoAcidSet.GetAminoAcid('T').Mass - 12.0238 - Composition.Hydrogen.Mass } } // for threonine
};
W = new BaseIonType("w", false, new CompositionWithDeltaMass(73.0290), aminoAcid => DefaultCompositionCalculator(wIonOffsets, aminoAcid));
AllBaseIonTypes = new List <BaseIonType> {
A, Ar, B, C, D, V, W, X, Xr, Y, YM1, Z, Zr
};
}
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();
* }
* } */
}
private Sequence LoadSequence(string sequenceStr, AminoAcidSet aminoAcidSet)
{
var sequence = new List <AminoAcid>();
for (int i = 0; i < sequenceStr.Length; i++)
{
SequenceLocation location;
if (i == 0)
{
location = SequenceLocation.ProteinNTerm;
}
else if (i == sequenceStr.Length - 1)
{
location = SequenceLocation.ProteinCTerm;
}
else
{
location = SequenceLocation.Everywhere;
}
var aa = aminoAcidSet.GetAminoAcid(sequenceStr[i], location);
sequence.Add(aa);
}
return(new Sequence(sequence));
}
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);
}
public void TestITraqMod()
{
var aminoAcidSet = new AminoAcidSet();
var p = aminoAcidSet.GetAminoAcid('P');
var a = aminoAcidSet.GetAminoAcid('A');
var q = aminoAcidSet.GetAminoAcid('Q');
var itraqMod = Modification.Itraq4Plex;
Console.WriteLine(itraqMod.Mass);
var modp = new ModifiedAminoAcid(p, itraqMod);
var sequence = new Sequence(new List <AminoAcid> {
modp, a, q
});
Console.WriteLine(sequence.Mass);
}
public static Sequence GenerateSequence(String seqStr, String modStr)
{
if (modStr == null || modStr.Equals(""))
{
return(new Sequence(seqStr, aaSet));
}
int residueIndex = 0;
string[] ptms = modStr.Split(',');
var modificaitonInfo = new Dictionary <int, Modification>();
foreach (var ptm in ptms)
{
string[] ptmPair = ptm.Split(' ');
residueIndex = Int32.Parse(ptmPair[1]);
if (residueIndex == 0)
{
residueIndex++;
}
modificaitonInfo.Add(residueIndex, Modification.Get(ptmPair[0]));
}
var aaList = new List <AminoAcid>();
residueIndex = 1;
foreach (var residue in seqStr)
{
var aa = aaSet.GetAminoAcid(residue);
if (modificaitonInfo.ContainsKey(residueIndex))
{
aaList.Add(new ModifiedAminoAcid(aa, modificaitonInfo[residueIndex]));
}
else
{
aaList.Add(aa);
}
residueIndex++;
}
Sequence sequence = new Sequence(aaList);
return(sequence);
}
/// <summary>
/// Update all modifications in the sequence.
/// </summary>
public void UpdateModifications()
{
var aminoAcidSet = new AminoAcidSet();
for (int i = 0; i < this.Sequence.Count; i++)
{
var modAminoAcid = this.Sequence[i] as ModifiedAminoAcid;
if (modAminoAcid != null)
{
var modification = Modification.Get(modAminoAcid.Modification.Name);
if (modification != null)
{
this.Sequence[i] = new ModifiedAminoAcid(aminoAcidSet.GetAminoAcid(modAminoAcid.Residue), modification);
}
}
}
this.Sequence = new Sequence(this.Sequence);
}
/// <summary>
/// Parse a protein/peptide sequence in the LCMSSpectator style.
/// </summary>
/// <param name="sequence">The sequence as a string.</param>
/// <returns>The parsed sequence.</returns>
public Sequence Read(string sequence)
{
if (this.trimAnnotations)
{
var firstIndex = sequence.IndexOf('.');
if (firstIndex >= 0)
{
var index = Math.Min(firstIndex + 1, sequence.Length - 1);
sequence = sequence.Substring(index, sequence.Length - index - 1);
}
var lastIndex = sequence.LastIndexOf('.');
if (lastIndex >= 0)
{
var index = Math.Min(lastIndex, sequence.Length - 1);
sequence = sequence.Substring(0, index);
}
}
const string AminoAcidRegex = @"[" + AminoAcid.StandardAminoAcidCharacters + "]";
////const string modRegex = @"\[([A-Z]|[a-z])+\]";
const string ModRegex = @"\[([A-Z]|[a-z]|[0-9]|-|>)+\]";
if (string.IsNullOrEmpty(sequence))
{
return(new Sequence(new List <AminoAcid>()));
}
if (!Regex.IsMatch(sequence, "(" + AminoAcidRegex + "|" + ModRegex + ")+"))
{
return(null);
}
var stdAaSet = new AminoAcidSet();
var aminoAcidList = new List <AminoAcid>();
var matches = Regex.Matches(sequence, "(" + AminoAcidRegex + "|" + ModRegex + ")");
AminoAcid aa = null;
var mods = new List <Modification>();
foreach (Match match in matches)
{
var element = match.Value;
if (element.Length == 0)
{
continue;
}
if (element.Length == 1 && char.IsLetter(element[0]))
{ // amino acid
if (aa != null)
{
aa = mods.Aggregate(aa, (current, mod) => new ModifiedAminoAcid(current, mod));
aminoAcidList.Add(aa);
mods = new List <Modification>();
}
aa = stdAaSet.GetAminoAcid(element[0]);
if (aa == null)
{
throw new FormatException("Unrecognized amino acid character: " + element[0]);
}
//// Console.WriteLine("{0} {1} {2}", aa.Residue, aa.Composition, aa.GetMass());
}
else
{
var modName = element.Substring(1, element.Length - 2);
var mod = Modification.Get(modName);
if (mod == null)
{
throw new FormatException("Unrecognized modification: " + modName);
}
mods.Add(mod);
//// Console.WriteLine("{0} {1} {2}", mod.Name, mod.Composition, mod.Composition.AveragineMass);
}
}
if (aa != null)
{
aa = mods.Aggregate(aa, (current, mod) => new ModifiedAminoAcid(current, mod));
aminoAcidList.Add(aa);
}
return(new Sequence(aminoAcidList));
}
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>
/// Build the data graph from a sequence graph.
/// </summary>
private void BuildGraph()
{
var sequenceRev = _sequence.Reverse();
var sequence = sequenceRev.Aggregate(string.Empty, (current, aa) => current + aa);
sequence = "\0" + sequence;
vertices = new DataVertex[_maxSeqIndex][];
var mods = AminoAcidSet.GetModificationParams();
var id = 0;
var ntermIndex = 0;
var start = _maxSeqIndex - 2;
var end = 0;
var offset = 1;
if (nterminal != null)
{
start++;
sequence += nterminal.Residue;
}
if (cterminal != null)
{
end--;
offset = 0;
sequence = sequence.Insert(1, cterminal.Residue.ToString(CultureInfo.InvariantCulture));
}
// create vertices
for (var si = start; si > end; si--)
{
var graphSi = si - offset;
vertices[graphSi] = new DataVertex[_graph[si].Length];
for (var mi = 0; mi < _graph[si].Length; mi++)
{
var node = _graph[si][mi];
var mod = mods.GetModificationCombination(node.ModificationCombinationIndex);
SetSink(mi);
vertices[graphSi][mi] = new DataVertex
{
ID = id++,
NTermIndex = ntermIndex,
ModIndex = mi,
PrefixComposition = GetComplementaryComposition(si, mi),
SuffixComposition = GetComposition(si, mi),
ModificationCombination = mod,
Text = string.Empty
};
var vertex = vertices[graphSi][mi];
DataGraph.AddVertex(vertex);
}
ntermIndex++;
}
// connect vertices
for (var si = start; si > (end + 1); si--)
{
var graphSi = si - offset;
for (var mi = 0; mi < _graph[si].Length; mi++)
{
var node = _graph[si][mi];
var currVertex = vertices[graphSi][mi];
foreach (var nextModIndex in node.GetPrevNodeIndices())
{
var nextVertex = vertices[graphSi - 1][nextModIndex];
var currVertexMods = currVertex.ModificationCombination.Modifications;
var nextVertexMods = nextVertex.ModificationCombination.Modifications;
var result = new List <Modification>(currVertexMods);
foreach (var mod in nextVertexMods)
{
if (result.Contains(mod))
{
result.Remove(mod);
}
}
AminoAcid aminoAcid;
if (si == start && nterminal != null)
{
aminoAcid = nterminal;
}
else if (si == end + 2 && cterminal != null)
{
aminoAcid = cterminal;
}
else
{
aminoAcid = AminoAcidSet.GetAminoAcid(sequence[graphSi]);
}
var modAa = aminoAcid as ModifiedAminoAcid;
Modification aminoAcidMod = null;
if (modAa != null)
{
aminoAcidMod = modAa.Modification;
}
if (aminoAcidMod != null)
{
result.Add(aminoAcidMod);
}
var edgeModifications = new ModificationCombination(result);
var edge = new DataEdge(currVertex, nextVertex)
{
AminoAcid = aminoAcid,
SequenceIndex = graphSi,
Modifications = edgeModifications
};
DataGraph.AddEdge(edge);
}
}
}
}