/// <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>
/// Parse the provided modification line
/// </summary>
/// <param name="line"></param>
/// <returns></returns>
public static List <SearchModification> ParseModification(string line)
{
var token = line.Split(',');
if (token.Length != 5)
{
return(null);
}
// Composition
var compStr = token[0].Trim();
var composition = Composition.Composition.ParseFromPlainString(compStr) ??
Composition.Composition.Parse(compStr);
if (composition == null)
{
throw new Exception(string.Format("Illegal Composition: \"{0}\" in \"{1}\"", compStr, line));
}
// Residues
var residueStr = token[1].Trim();
var isResidueStrLegitimate = residueStr.Equals("*") ||
residueStr.Any() && residueStr.All(AminoAcid.IsStandardAminoAcidResidue);
if (!isResidueStrLegitimate)
{
throw new Exception(string.Format("Illegal residues: \"{0}\" in \"{1}\"", residueStr, line));
}
// isFixedModification
bool isFixedModification;
if (token[2].Trim().Equals("fix", StringComparison.InvariantCultureIgnoreCase))
{
isFixedModification = true;
}
else if (token[2].Trim().Equals("opt", StringComparison.InvariantCultureIgnoreCase))
{
isFixedModification = false;
}
else
{
throw new Exception(string.Format("Illegal modification type (fix or opt): \"{0}\" in \"{1}\"", token[2].Trim(), line));
}
// Location
SequenceLocation location;
var locStr = token[3].Trim().Split()[0];
if (locStr.Equals("any", StringComparison.InvariantCultureIgnoreCase))
{
location = SequenceLocation.Everywhere;
}
else if (locStr.Equals("N-Term", StringComparison.InvariantCultureIgnoreCase) ||
locStr.Equals("NTerm", StringComparison.InvariantCultureIgnoreCase))
{
location = SequenceLocation.PeptideNTerm;
}
else if (locStr.Equals("C-Term", StringComparison.InvariantCultureIgnoreCase) ||
locStr.Equals("CTerm", StringComparison.InvariantCultureIgnoreCase))
{
location = SequenceLocation.PeptideCTerm;
}
else if (locStr.Equals("Prot-N-Term", StringComparison.InvariantCultureIgnoreCase) ||
locStr.Equals("ProtNTerm", StringComparison.InvariantCultureIgnoreCase))
{
location = SequenceLocation.ProteinNTerm;
}
else if (locStr.Equals("Prot-C-Term", StringComparison.InvariantCultureIgnoreCase) ||
locStr.Equals("ProtCTerm", StringComparison.InvariantCultureIgnoreCase))
{
location = SequenceLocation.ProteinCTerm;
}
else
{
throw new Exception(string.Format("Illegal modification location (any|(Prot-?)?(N|C)-?Term): \"{0}\" in \"{1}\"", token[3].Trim(), line));
}
// Check if it's valid
if (residueStr.Equals("*") && location == SequenceLocation.Everywhere)
{
throw new Exception(string.Format("Invalid modification: * should not be applied to \"any\": \"{0}\"", line));
}
var name = token[4].Split()[0].Trim();
var mod = Modification.Get(name) ?? Modification.RegisterAndGetModification(name, composition);
return(residueStr.Select(residue => new SearchModification(mod, residue, location, isFixedModification)).ToList());
}
private static IEnumerable <SearchModification> Parse(string modFilePath, out int maxNumDynModsPerPeptide)
{
var searchModList = new List <SearchModification>();
var numMods = 0;
var lineNum = 0;
foreach (var line in File.ReadLines(modFilePath))
{
lineNum++;
var tokenArr = line.Split('#');
if (tokenArr.Length == 0)
{
continue;
}
var s = tokenArr[0].Trim();
if (s.Length == 0)
{
continue;
}
if (s.StartsWith("NumMods="))
{
try {
numMods = Convert.ToInt32(s.Split('=')[1].Trim());
} catch (FormatException)
{
Console.WriteLine("{0}: Illegal NumMods parameter at line {1} - {2}", modFilePath, lineNum, s);
maxNumDynModsPerPeptide = -1;
return(null);
}
}
else
{
var token = s.Split(',');
if (token.Length != 5)
{
continue;
}
// Composition
var compStr = token[0].Trim();
var composition = Composition.Composition.ParseFromPlainString(compStr) ??
Composition.Composition.Parse(compStr);
if (composition == null)
{
Console.WriteLine("{0}: Illegal Composition at line {1} - {2}", modFilePath, lineNum, s);
maxNumDynModsPerPeptide = -1;
return(null);
}
// Residues
var residueStr = token[1].Trim();
var isResidueStrLegitimate = residueStr.Equals("*") ||
residueStr.Any() && residueStr.All(AminoAcid.IsStandardAminoAcidResidue);
if (!isResidueStrLegitimate)
{
Console.WriteLine("{0}: Illegal residues at line {1} - {2}", modFilePath, lineNum, s);
maxNumDynModsPerPeptide = -1;
return(null);
}
// isFixedModification
bool isFixedModification;
if (token[2].Trim().Equals("fix", StringComparison.InvariantCultureIgnoreCase))
{
isFixedModification = true;
}
else if (token[2].Trim().Equals("opt", StringComparison.InvariantCultureIgnoreCase))
{
isFixedModification = false;
}
else
{
Console.WriteLine("{0}: Illegal modification type (fix or opt) at line {1} - {2}", modFilePath, lineNum, s);
maxNumDynModsPerPeptide = -1;
return(null);
}
// Location
SequenceLocation location;
var locStr = token[3].Trim().Split()[0];
if (locStr.Equals("any", StringComparison.InvariantCultureIgnoreCase))
{
location = SequenceLocation.Everywhere;
}
else if (locStr.Equals("N-Term", StringComparison.InvariantCultureIgnoreCase) ||
locStr.Equals("NTerm", StringComparison.InvariantCultureIgnoreCase))
{
location = SequenceLocation.PeptideNTerm;
}
else if (locStr.Equals("C-Term", StringComparison.InvariantCultureIgnoreCase) ||
locStr.Equals("CTerm", StringComparison.InvariantCultureIgnoreCase))
{
location = SequenceLocation.PeptideCTerm;
}
else if (locStr.Equals("Prot-N-Term", StringComparison.InvariantCultureIgnoreCase) ||
locStr.Equals("ProtNTerm", StringComparison.InvariantCultureIgnoreCase))
{
location = SequenceLocation.ProteinNTerm;
}
else if (locStr.Equals("Prot-C-Term", StringComparison.InvariantCultureIgnoreCase) ||
locStr.Equals("ProtCTerm", StringComparison.InvariantCultureIgnoreCase))
{
location = SequenceLocation.ProteinCTerm;
}
else
{
Console.WriteLine("{0}: Illegal modification location (fix or opt) at line {1} - {2}", modFilePath, lineNum, s);
maxNumDynModsPerPeptide = -1;
return(null);
}
// Check if it's valid
if (residueStr.Equals("*") && location == SequenceLocation.Everywhere)
{
Console.WriteLine("{0}: Invalid modification: * should not be applied to \"any\"", modFilePath);
maxNumDynModsPerPeptide = -1;
return(null);
}
var name = token[4].Split()[0].Trim();
var mod = Modification.Get(name) ?? Modification.RegisterAndGetModification(name, composition);
searchModList.AddRange(residueStr.Select(
residue => new SearchModification(mod, residue, location, isFixedModification)
));
}
}
maxNumDynModsPerPeptide = numMods;
return(searchModList);
}