private IEnumerable <FlankingMassMatch> GetBackwardMatches(
MatchedTag matchedTag,
ShiftedSequenceGraph backwardGraph,
double?featureMass = null
)
{
for (var j = matchedTag.StartIndex - 1; j >= -1; j--)
{
var residue = j >= 0 ? _proteinSequence[j] : AminoAcid.ProteinNTerm.Residue;
var location = j > 0 ? SequenceLocation.Everywhere : SequenceLocation.ProteinNTerm;
if (!backwardGraph.AddAminoAcid(residue, location))
{
yield break;
}
if (j == 0)
{
continue;
}
var backwardMatch = GetBestMatchInTheGraph(backwardGraph, _spec, featureMass);
if (backwardMatch != null)
{
backwardMatch.Index = Math.Max(j, 0);
yield return(backwardMatch);
}
}
}
// private readonly int _minProductIonCharge;
// private readonly int _maxProductIonCharge;
private IEnumerable <FlankingMassMatch> GetForwardMatches(
MatchedTag matchedTag,
ShiftedSequenceGraph forwardGraph,
double?featureMass = null
)
{
for (var i = matchedTag.EndIndex; i <= _proteinSequence.Length; i++)
{
var residue = i < _proteinSequence.Length ? _proteinSequence[i] : AminoAcid.ProteinCTerm.Residue;
var location = i < _proteinSequence.Length - 1
? SequenceLocation.Everywhere
: SequenceLocation.ProteinCTerm;
if (!forwardGraph.AddAminoAcid(residue, location))
{
yield break;
}
if (i == _proteinSequence.Length - 1)
{
continue;
}
var forwardMatch = GetBestMatchInTheGraph(forwardGraph, _spec, featureMass);
if (forwardMatch != null)
{
forwardMatch.Index = Math.Min(i + 1, _proteinSequence.Length);
yield return(forwardMatch);
}
}
}
private IEnumerable <TagMatch> FindMatchesWithFeatureMass(MatchedTag matchedTag)
{
if (matchedTag.NTermFlankingMass == null || matchedTag.CTermFlankingMass == null)
{
yield break;
}
var featureMass = (double)matchedTag.NTermFlankingMass + matchedTag.Mass +
(double)matchedTag.CTermFlankingMass + Composition.H2O.Mass;
var shiftMass = matchedTag.Mass + (double)matchedTag.NTermFlankingMass;
var backwardGraph = new ShiftedSequenceGraph(_aaSet, shiftMass, false,
matchedTag.StartIndex, featureMass - MinSumModificationMasses);
foreach (var backwardMatch in GetBackwardMatches(matchedTag, backwardGraph, featureMass))
{
// Make a forward graph
var nTermShiftMass = backwardMatch.Mass + matchedTag.Mass;
var forwardGraph = new ShiftedSequenceGraph(_aaSet, nTermShiftMass, true,
_proteinSequence.Length - matchedTag.EndIndex, featureMass - MinSumModificationMasses);
foreach (
var forwardMatch in
GetForwardMatches(matchedTag, forwardGraph, featureMass))
{
var mass = forwardMatch.Mass + matchedTag.Mass + backwardMatch.Mass;
if (mass > _maxSequenceMass)
{
continue;
}
var offset = matchedTag.EndIndex - backwardMatch.Index - 1;
var modStr = string.Join(",", backwardMatch.Modifications.Concat(forwardMatch.Modifications.Select(m => m.GetModificationInstanceWithOffset(offset))));
var modList = new List <Modification>();
foreach (var mod in backwardMatch.Modifications)
{
modList.Add(mod.Modification);
}
foreach (var mod in forwardMatch.Modifications)
{
modList.Add(mod.Modification);
}
var tagMatch = new TagMatch(
backwardMatch.Index,
forwardMatch.Index,
matchedTag.Length,
backwardMatch.Charge,
backwardMatch.Score,
forwardMatch.Score,
mass,
new ModificationCombination(modList),
modStr);
yield return(tagMatch);
}
}
}
public static Dictionary <string, MatchedTagSet> GetProteinToMatchedTagsMap(
IEnumerable <SequenceTag.SequenceTag> tags,
SearchableDatabase searchableDb,
AminoAcidSet aaSet,
Tolerance tolerance,
Tolerance relaxedTolerance)
{
var fastaDb = searchableDb.FastaDatabase;
var proteinsToTags = new Dictionary <string, MatchedTagSet>();
foreach (var tag in tags)
{
var matchedIndices = searchableDb.FindAllMatchedSequenceIndices(tag.Sequence).ToArray();
if (matchedIndices.Length > MaxNumProteinMatchesPerTag)
{
continue;
}
foreach (var index in matchedIndices)
{
var proteinName = fastaDb.GetProteinName(index);
var startIndex = fastaDb.GetZeroBasedPositionInProtein(index);
var mass = aaSet.GetComposition(tag.Sequence).Mass;
var matchedTag = new MatchedTag(tag, startIndex)
{
Mass = mass
};
MatchedTagSet existingMatchedTagSet;
if (proteinsToTags.TryGetValue(proteinName, out existingMatchedTagSet))
{
existingMatchedTagSet.Add(matchedTag);
}
else
{
var proteinSequence = fastaDb.GetProteinSequence(proteinName);
if (proteinSequence == null)
{
proteinSequence = proteinName;
}
var matchedTagSet = new MatchedTagSet(proteinSequence, aaSet, tolerance, relaxedTolerance);
matchedTagSet.Add(matchedTag);
proteinsToTags.Add(proteinName, matchedTagSet);
}
}
}
return(proteinsToTags);
}
public IEnumerable <TagMatch> FindMatches(MatchedTag matchedTag)
{
if (matchedTag.NTermFlankingMass != null && matchedTag.CTermFlankingMass != null)
{
return(FindMatchesWithFeatureMass(matchedTag));
}
if (matchedTag.NTermFlankingMass != null)
{
return(FindMatchesForwardAndBackward(matchedTag));
}
if (matchedTag.CTermFlankingMass != null)
{
return(FindMatchesBackwardAndForward(matchedTag));
}
return(Enumerable.Empty <TagMatch>());
}
public void TestFeatureId()
{
var methodName = MethodBase.GetCurrentMethod().Name;
TestUtils.ShowStarting(methodName);
const string dataSet = @"H:\Research\QCShew_TopDown\Production\QC_Shew_Intact_26Sep14_Bane_C2Column3";
if (!File.Exists(dataSet))
{
Assert.Ignore(@"Skipping test {0} since file not found: {1}", methodName, dataSet);
}
// Feature: 5236-5286 6-12 8480.3681 5
const int minScanNum = 5236;
const int maxScanNum = 5286;
const double featureMass = 8480.3681;
//const int minScanNum = 7251;
//const int maxScanNum = 7326;
//const double featureMass = 32347.18;
// const int minScanNum = 4451;
// const int maxScanNum = 4541;
// const double featureMass = 31267.95;
var tolerance = new Tolerance(10);
var relaxedTolerance = new Tolerance(20);
const int minTagLength = 5;
const int minMergedTagLength = 7;
const int minNumTagMatches = 1;
var rawFileName = MassSpecDataReaderFactory.ChangeExtension(dataSet, ".raw");
var run = PbfLcMsRun.GetLcMsRun(rawFileName);
var aminoAcidSet = AminoAcidSet.GetStandardAminoAcidSet();
var featureFileName = MassSpecDataReaderFactory.ChangeExtension(dataSet, ".ms1ft");
var filter = new Ms1FtFilter(run, tolerance, featureFileName);
var ms2ScanNums =
filter.GetMatchingMs2ScanNums(featureMass)
.Where(scanNum => scanNum > minScanNum && scanNum < maxScanNum)
.ToArray();
const string tagFileName = dataSet + ".seqtag"; //"_MinLength3.seqtag"; //Path.ChangeExtension(dataSet, ".seqtag");
const string fastaFilePath = @"H:\Research\QCShew_TopDown\Production\ID_002216_235ACCEA.fasta";
if (!File.Exists(fastaFilePath))
{
Assert.Ignore(@"Skipping test {0} since file not found: {1}", methodName, fastaFilePath);
}
var fastaDb = new FastaDatabase(fastaFilePath);
var searchableDb = new SearchableDatabase(fastaDb);
var tagParser = new SequenceTagParser(tagFileName, minTagLength);
var proteinsToTags = new Dictionary <string, IList <MatchedTag> >();
foreach (var ms2ScanNum in ms2ScanNums)
{
var tags = tagParser.GetSequenceTags(ms2ScanNum);
foreach (var tag in tags)
{
var matchedIndices = searchableDb.FindAllMatchedSequenceIndices(tag.Sequence).ToArray();
foreach (var index in matchedIndices)
{
var protein = fastaDb.GetProteinName(index);
var startIndex = fastaDb.GetZeroBasedPositionInProtein(index);
var matchedTag = new MatchedTag(tag, startIndex, featureMass);
IList <MatchedTag> existingTags;
if (proteinsToTags.TryGetValue(protein, out existingTags))
{
existingTags.Add(matchedTag);
}
else
{
proteinsToTags.Add(protein, new List <MatchedTag> {
matchedTag
});
}
}
}
}
foreach (var entry in proteinsToTags.OrderByDescending(e => e.Value.Count))
{
if (entry.Value.Count < minNumTagMatches)
{
break;
}
var proteinName = entry.Key;
var proteinSequence = fastaDb.GetProteinSequence(proteinName);
var protein = new Sequence(proteinSequence, aminoAcidSet);
Console.WriteLine(proteinName + "\t" + entry.Value.Count);
var matchedTagSet = new MatchedTagSet(proteinSequence, aminoAcidSet,
tolerance, relaxedTolerance);
Console.WriteLine("********** Before merging");
foreach (var matchedTag in entry.Value)
{
var seq = proteinSequence.Substring(matchedTag.StartIndex,
matchedTag.EndIndex - matchedTag.StartIndex);
var nTermMass = protein.GetMass(0, matchedTag.StartIndex);
var cTermMass = protein.GetMass(matchedTag.EndIndex, protein.Count);
Console.WriteLine("\t{0}\t{1}\t{2}\t{3}\t{4}\t{5}",
(matchedTag.NTermFlankingMass - nTermMass), seq, (matchedTag.CTermFlankingMass - cTermMass), matchedTag.StartIndex,
matchedTag.IsNTermFlankingMassReliable, matchedTag.IsCTermFlankingMassReliable);
matchedTagSet.Add(matchedTag);
}
Console.WriteLine("********** After merging");
foreach (var matchedTag in matchedTagSet.Tags)
{
if (matchedTag.Length < minMergedTagLength)
{
continue;
}
var seq = proteinSequence.Substring(matchedTag.StartIndex,
matchedTag.EndIndex - matchedTag.StartIndex);
var nTermMass = protein.GetMass(0, matchedTag.StartIndex);
var cTermMass = protein.GetMass(matchedTag.EndIndex, protein.Count);
Console.WriteLine("\t{0}\t{1}\t{2}\t{3}\t{4}\t{5}",
(matchedTag.NTermFlankingMass - nTermMass), seq, (matchedTag.CTermFlankingMass - cTermMass), matchedTag.StartIndex,
matchedTag.IsNTermFlankingMassReliable, matchedTag.IsCTermFlankingMassReliable);
}
break;
}
}
public void TestGetProteinsWithTagMatchingSingleSpec()
{
var methodName = MethodBase.GetCurrentMethod().Name;
TestUtils.ShowStarting(methodName);
const string dataSet = @"H:\Research\Lewy\raw\Lewy_intact_07";
// const int scanNum = 5158;
const int minTagLength = 7;
const int minNumTagMatches = 1;
var aminoAcidSet = AminoAcidSet.GetStandardAminoAcidSet();
const int scanNum = 2;
// Parse sequence tags
//const string tagFileName = dataSet + ".seqtag"; //"_MinLength3.seqtag"; //Path.ChangeExtension(dataSet, ".seqtag");
const string rawFilePath = "";
const string fastaFilePath = @"H:\Research\Lewy\ID_004858_0EE8CF61.fasta";
if (!File.Exists(rawFilePath))
{
Assert.Ignore(@"Skipping test {0} since file not found: {1}", methodName, rawFilePath);
}
if (!File.Exists(fastaFilePath))
{
Assert.Ignore(@"Skipping test {0} since file not found: {1}", methodName, fastaFilePath);
}
var fastaDb = new FastaDatabase(fastaFilePath);
var searchableDb = new SearchableDatabase(fastaDb);
//var tagParser = new SequenceTagParser(tagFileName, minTagLength);
//var tags = tagParser.GetSequenceTags(scanNum);
var run = PbfLcMsRun.GetLcMsRun(rawFilePath);
var spec = run.GetSpectrum(scanNum) as ProductSpectrum;
var tagFinder = new SequenceTagFinder(spec, new Tolerance(5));
var tags = tagFinder.GetAllSequenceTagString();
var proteinsToTags = new Dictionary <string, IList <MatchedTag> >();
foreach (var tag in tags)
{
var matchedIndices = searchableDb.FindAllMatchedSequenceIndices(tag.Sequence).ToArray();
foreach (var index in matchedIndices)
{
var protein = fastaDb.GetProteinName(index);
var startIndex = fastaDb.GetOneBasedPositionInProtein(index);
var matchedTag = new MatchedTag(tag, startIndex, 0.0);
IList <MatchedTag> existingTags;
if (proteinsToTags.TryGetValue(protein, out existingTags))
{
existingTags.Add(matchedTag);
}
else
{
proteinsToTags.Add(protein, new List <MatchedTag> {
matchedTag
});
}
}
}
foreach (var entry in proteinsToTags.OrderByDescending(e => e.Value.Count))
{
if (entry.Value.Count < minNumTagMatches)
{
break;
}
var proteinName = entry.Key;
var proteinSequence = fastaDb.GetProteinSequence(proteinName);
var protein = new Sequence(proteinSequence, aminoAcidSet);
Console.WriteLine(proteinName + "\t" + entry.Value.Count);
foreach (var matchedTag in entry.Value)
{
var seq = proteinSequence.Substring(matchedTag.StartIndex,
matchedTag.EndIndex - matchedTag.StartIndex);
var nTermMass = protein.GetMass(0, matchedTag.StartIndex);
var cTermMass = protein.GetMass(matchedTag.EndIndex, protein.Count);
Console.WriteLine("\t{0} ({1})\t{2}\t{3} ({4})\t{5}\t{6}\t{7}",
matchedTag.NTermFlankingMass, (matchedTag.NTermFlankingMass - nTermMass),
seq,
matchedTag.CTermFlankingMass, (matchedTag.CTermFlankingMass - cTermMass),
matchedTag.StartIndex,
matchedTag.IsNTermFlankingMassReliable, matchedTag.IsCTermFlankingMassReliable);
}
}
}
