public ProteinScoringGraphFactory(IMassBinning comparer, AminoAcidSet aminoAcidSet)
{
_comparer = comparer;
_adjList = new LinkedList<ScoringGraphEdge>[_comparer.NumberOfBins];
for (var i = 0; i < _comparer.NumberOfBins; i++) _adjList[i] = new LinkedList<ScoringGraphEdge>();
var terminalModifications = FilteredProteinMassBinning.GetTerminalModifications(aminoAcidSet);
var aminoAcidArray = FilteredProteinMassBinning.GetExtendedAminoAcidArray(aminoAcidSet);
for (var i = 0; i < _comparer.NumberOfBins; i++)
{
var mi = _comparer.GetMass(i);
var fineNodeMass = mi;
for (var a = 0; a < aminoAcidArray.Length; a++)
{
var aa = aminoAcidArray[a];
var j = _comparer.GetBinNumber(fineNodeMass + aa.Mass);
if (j < 0 || j >= _comparer.NumberOfBins) continue;
_adjList[j].AddLast(new ScoringGraphEdge(i));
if (i == 0 && !(aa is ModifiedAminoAcid))
{
foreach (var terminalMod in terminalModifications)
{
var modifiedAa = new ModifiedAminoAcid(aa, terminalMod);
j = _comparer.GetBinNumber(fineNodeMass + modifiedAa.Mass);
if (j < 0 || j >= _comparer.NumberOfBins) continue;
_adjList[j].AddLast(new ScoringGraphEdge(i));
}
}
}
}
}
// Generate an amino acid map with Cys static modification
public AminoAcidSet(Modification cysMod) : this()
{
foreach (var loc in AllSequenceLocations)
{
var residueMap = _locationSpecificResidueMap[loc];
residueMap['C'] = new ModifiedAminoAcid(residueMap['C'], cysMod);
}
}
// Generate an amino acid map with Cys static modification
public AminoAcidSet(Modification cysMod): this()
{
foreach (var loc in AllSequenceLocations)
{
var residueMap = _locationSpecificResidueMap[loc];
residueMap['C'] = new ModifiedAminoAcid(residueMap['C'], cysMod);
}
}
/// <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));
}
public Composition.Composition[] GetUniqueCompositions()
{
//var aaList = AminoAcid.StandardAminoAcidArr.ToList();
var uniqueCompositionList = new List <Composition.Composition>();
var tempModTable = new List <Composition.Composition> [AminoAcid.StandardAminoAcidArr.Length];
for (var i = 0; i < AminoAcid.StandardAminoAcidArr.Length; i++)
{
tempModTable[i] = new List <Composition.Composition>();
uniqueCompositionList.Add(AminoAcid.StandardAminoAcidArr[i].Composition);
}
foreach (var location in AllSequenceLocations)
{
//var keys = _locationSpecificResidueMap[location].Keys.ToArray();
for (var i = 0; i < AminoAcid.StandardAminoAcidArr.Length; i++)
{
var residue = AminoAcid.StandardAminoAcidArr[i].Residue;
var aa = GetAminoAcid(residue, location);
foreach (var modIndex in GetModificationIndices(residue, location))
{
var mAa = new ModifiedAminoAcid(aa, _modificationParams.GetModification(modIndex));
if (tempModTable[i].Contains(mAa.Composition))
{
continue;
}
tempModTable[i].Add(mAa.Composition);
uniqueCompositionList.Add(mAa.Composition);
//aaList.Add(mAa);
}
//Console.WriteLine();
}
}
return(uniqueCompositionList.ToArray());
}
public FilteredProteinMassBinning(AminoAcidSet aaSet, double maxProteinMass = 50000, int numBits = 27)
{
_aminoAcidSet = aaSet;
var terminalModifications = GetTerminalModifications(aaSet);
var extendedAminoAcidArray = GetExtendedAminoAcidArray(aaSet);
MaxMass = maxProteinMass;
MinMass = MaxMass;
foreach (var aa in extendedAminoAcidArray)
{
if (aa.Mass < MinMass) MinMass = aa.Mass;
foreach (var mod in terminalModifications)
{
var modAa = new ModifiedAminoAcid(aa, mod);
if (modAa.Mass < MinMass) MinMass = modAa.Mass;
}
}
_mzComparer = new MzComparerWithBinning(numBits);
_minMzBinIndex = _mzComparer.GetBinNumber(MinMass);
_maxMzBinIndex = _mzComparer.GetBinNumber(MaxMass);
var numberOfMzBins = _maxMzBinIndex - _minMzBinIndex + 2; // pad zero mass bin
_mzBinToFilteredBinMap = new int[numberOfMzBins];
for (var i = 0; i < numberOfMzBins; i++) _mzBinToFilteredBinMap[i] = -1;
var tempMap = new int[numberOfMzBins];
var fineNodes = new BitArray(Constants.GetBinNumHighPrecision(MaxMass));
fineNodes[0] = true;
var effectiveBinCounter = 0;
for (var fineBinIdx = 0; fineBinIdx < fineNodes.Length; fineBinIdx++)
{
if (!fineNodes[fineBinIdx]) continue;
var fineNodeMass = fineBinIdx / Constants.RescalingConstantHighPrecision;
foreach (var aa in extendedAminoAcidArray)
{
var validFineNodeIndex = Constants.GetBinNumHighPrecision(fineNodeMass + aa.Mass);
if (validFineNodeIndex >= fineNodes.Length) break;
fineNodes[validFineNodeIndex] = true;
if (fineBinIdx == 0 && !(aa is ModifiedAminoAcid)) // include terminal modifications
{
foreach (var terminalMod in terminalModifications)
{
var modifiedAa = new ModifiedAminoAcid(aa, terminalMod);
validFineNodeIndex = Constants.GetBinNumHighPrecision(fineNodeMass + modifiedAa.Mass);
if (validFineNodeIndex >= fineNodes.Length) break;
fineNodes[validFineNodeIndex] = true;
}
}
}
/*foreach (var m in massList)
{
var validFineNodeIndex = Constants.GetBinNumHighPrecision(fineNodeMass + m);
if (validFineNodeIndex >= fineNodes.Length) break;
fineNodes[validFineNodeIndex] = true;
}*/
var binNum = _mzComparer.GetBinNumber(fineNodeMass);
if (fineBinIdx == 0 || (binNum >= _minMzBinIndex && binNum <= _maxMzBinIndex && _mzBinToFilteredBinMap[binNum - _minMzBinIndex + 1] < 0))
{
_mzBinToFilteredBinMap[binNum == 0 ? 0 : binNum - _minMzBinIndex + 1] = effectiveBinCounter;
tempMap[effectiveBinCounter] = binNum;
effectiveBinCounter++;
}
}
_filteredBinToMzBinMap = new int[effectiveBinCounter];
Array.Copy(tempMap, _filteredBinToMzBinMap, effectiveBinCounter);
}
public static AminoAcid[] GetExtendedAminoAcidArray(AminoAcidSet aaSet)
{
var ret = new List<AminoAcid>();
var modParam = aaSet.GetModificationParams();
var aminoAcidArray = AminoAcid.StandardAminoAcidArr;
foreach (var aa in aminoAcidArray)
{
ret.Add(aa);
foreach (var modIndex in aaSet.GetModificationIndices(aa.Residue, SequenceLocation.Everywhere))
{
var aa2 = new ModifiedAminoAcid(aa, modParam.GetModification(modIndex));
ret.Add(aa2);
}
}
return ret.ToArray();
}
public AminoAcidSet(IEnumerable <SearchModification> searchModifications, int maxNumModsPerSequence) : this()
{
if (searchModifications == null)
{
return;
}
var modifications = searchModifications as SearchModification[] ?? searchModifications.ToArray();
// apply fixed modifications
foreach (var searchModification in modifications)
{
if (!searchModification.IsFixedModification)
{
continue;
}
var location = searchModification.Location;
var targetResidue = searchModification.TargetResidue;
foreach (var loc in AffectedLocations[location])
{
var appliedResidue = targetResidue != '*' ? targetResidue : SequenceLocationToLocationResidue[loc];
var residueMap = _locationSpecificResidueMap[loc];
residueMap[appliedResidue] = new ModifiedAminoAcid(residueMap[appliedResidue], searchModification.Modification);
}
}
if (maxNumModsPerSequence <= 0)
{
return;
}
// apply dynamic modifications
var dynamicModifications = new HashSet <Modification>();
var locationSpecificResidueVariableModMap = new Dictionary <SequenceLocation, Dictionary <char, List <Modification> > >();
foreach (var loc in AllSequenceLocations)
{
locationSpecificResidueVariableModMap[loc] = new Dictionary <char, List <Modification> >();
}
foreach (var searchModification in modifications)
{
if (searchModification.IsFixedModification)
{
continue;
}
dynamicModifications.Add(searchModification.Modification);
var location = searchModification.Location;
var targetResidue = searchModification.TargetResidue;
foreach (var loc in AffectedLocations[location])
{
var residueModMap = locationSpecificResidueVariableModMap[loc];
var appliedResidue = targetResidue != '*' ? targetResidue : SequenceLocationToLocationResidue[loc];
List <Modification> modList;
if (residueModMap.TryGetValue(appliedResidue, out modList))
{
modList.Add(searchModification.Modification);
}
else
{
residueModMap.Add(appliedResidue, new List <Modification> {
searchModification.Modification
});
}
}
}
var dynModArray = dynamicModifications.ToArray();
_modificationParams = new ModificationParams(dynModArray, maxNumModsPerSequence);
foreach (var loc in AllSequenceLocations)
{
var residueModMap = locationSpecificResidueVariableModMap[loc];
foreach (var entry in residueModMap)
{
var residue = entry.Key;
var modList = entry.Value;
var modIndexArr = new int[modList.Count];
var index = -1;
foreach (var mod in modList)
{
var modIndex = Array.IndexOf(dynModArray, mod);
modIndexArr[++index] = modIndex;
}
_locationSpecificResidueModMap[loc].Add(residue, modIndexArr);
}
}
}
public AminoAcidSet(IEnumerable<SearchModification> searchModifications, int maxNumModsPerSequence): this()
{
if (searchModifications == null) return;
var modifications = searchModifications as SearchModification[] ?? searchModifications.ToArray();
// apply fixed modifications
foreach (var searchModification in modifications)
{
if (!searchModification.IsFixedModification) continue;
var location = searchModification.Location;
var targetResidue = searchModification.TargetResidue;
foreach (var loc in AffectedLocations[location])
{
var appliedResidue = targetResidue != '*' ? targetResidue : SequenceLocationToLocationResidue[loc];
var residueMap = _locationSpecificResidueMap[loc];
residueMap[appliedResidue] = new ModifiedAminoAcid(residueMap[appliedResidue], searchModification.Modification);
}
}
if (maxNumModsPerSequence <= 0) return;
// apply dynamic modifications
var dynamicModifications = new HashSet<Modification>();
var locationSpecificResidueVariableModMap = new Dictionary<SequenceLocation,Dictionary<char, List<Modification>>>();
foreach (var loc in AllSequenceLocations)
{
locationSpecificResidueVariableModMap[loc] = new Dictionary<char, List<Modification>>();
}
foreach (var searchModification in modifications)
{
if (searchModification.IsFixedModification) continue;
dynamicModifications.Add(searchModification.Modification);
var location = searchModification.Location;
var targetResidue = searchModification.TargetResidue;
foreach (var loc in AffectedLocations[location])
{
var residueModMap = locationSpecificResidueVariableModMap[loc];
var appliedResidue = targetResidue != '*' ? targetResidue : SequenceLocationToLocationResidue[loc];
List<Modification> modList;
if (residueModMap.TryGetValue(appliedResidue, out modList))
{
modList.Add(searchModification.Modification);
}
else
{
residueModMap.Add(appliedResidue, new List<Modification> { searchModification.Modification });
}
}
}
var dynModArray = dynamicModifications.ToArray();
_modificationParams = new ModificationParams(dynModArray, maxNumModsPerSequence);
foreach (var loc in AllSequenceLocations)
{
var residueModMap = locationSpecificResidueVariableModMap[loc];
foreach (var entry in residueModMap)
{
var residue = entry.Key;
var modList = entry.Value;
var modIndexArr = new int[modList.Count];
var index = -1;
foreach (var mod in modList)
{
var modIndex = Array.IndexOf(dynModArray, mod);
modIndexArr[++index] = modIndex;
}
_locationSpecificResidueModMap[loc].Add(residue, modIndexArr);
}
}
}
public Composition.Composition[] GetUniqueCompositions()
{
//var aaList = AminoAcid.StandardAminoAcidArr.ToList();
var uniqueCompositionList = new List<Composition.Composition>();
var tempModTable = new List<Composition.Composition>[AminoAcid.StandardAminoAcidArr.Length];
for (var i = 0; i < AminoAcid.StandardAminoAcidArr.Length; i++)
{
tempModTable[i] = new List<Composition.Composition>();
uniqueCompositionList.Add(AminoAcid.StandardAminoAcidArr[i].Composition);
}
foreach (var location in AllSequenceLocations)
{
//var keys = _locationSpecificResidueMap[location].Keys.ToArray();
for(var i =0; i < AminoAcid.StandardAminoAcidArr.Length; i++)
{
var residue = AminoAcid.StandardAminoAcidArr[i].Residue;
var aa = GetAminoAcid(residue, location);
foreach (var modIndex in GetModificationIndices(residue, location))
{
var mAa = new ModifiedAminoAcid(aa, _modificationParams.GetModification(modIndex));
if (tempModTable[i].Contains(mAa.Composition)) continue;
tempModTable[i].Add(mAa.Composition);
uniqueCompositionList.Add(mAa.Composition);
//aaList.Add(mAa);
}
//Console.WriteLine();
}
}
return uniqueCompositionList.ToArray();
}