public void TestIsMutationOneIgnoreDeamidated()
{
int site = -1;
//N->D
Assert.IsTrue(MutationUtils.IsMutationOne2("ABNDEIR", "ABDDEIR", ref site, false, false, false));
//N->D
Assert.IsFalse(MutationUtils.IsMutationOne2("ABNDEIR", "ABDDEIR", ref site, false, true, false));
//Q->E
Assert.IsTrue(MutationUtils.IsMutationOne2("ABQDEIR", "ABEDEIR", ref site, false, false, false));
//Q->E
Assert.IsFalse(MutationUtils.IsMutationOne2("ABQDEIR", "ABEDEIR", ref site, false, true, false));
//N->D
Assert.IsTrue(MutationUtils.IsMutationOneIL2("ABNDEIR", "ABDDELR", ref site, false, false, false));
//N->D
Assert.IsFalse(MutationUtils.IsMutationOneIL2("ABNDEIR", "ABDDELR", ref site, false, true, false));
//Q->E
Assert.IsTrue(MutationUtils.IsMutationOneIL2("ABQDEIR", "ABEDELR", ref site, false, false, false));
//Q->E
Assert.IsFalse(MutationUtils.IsMutationOneIL2("ABQDEIR", "ABEDELR", ref site, false, true, false));
}
public void TestIsAminoacidEquals()
{
Assert.IsTrue(MutationUtils.IsAminoacidEquals('I', 'L'));
Assert.IsTrue(MutationUtils.IsAminoacidEquals('L', 'I'));
Assert.IsFalse(MutationUtils.IsAminoacidEquals('L', 'A'));
}
private Dictionary <char, List <TargetVariant> > GetSubstitutionDeltaMass()
{
var result = new Dictionary <char, List <TargetVariant> >();
var aa = new Aminoacids();
var validAA = aa.GetVisibleAminoacids();
foreach (var ai in validAA)
{
foreach (var aj in validAA)
{
if (ai == aj)
{
continue;
}
if (this.IsSingleNucleotideMutationOnly && !MutationUtils.IsSingleNucleotideMutation(ai, aj))
{
continue;
}
if (this.IgnoreDeamidatedMutation && MutationUtils.IsDeamidatedMutation(ai, aj))
{
continue;
}
if (!result.ContainsKey(ai))
{
result[ai] = new List <TargetVariant>();
}
var deltaMass = aa[aj].MonoMass - aa[ai].MonoMass;
if (Math.Abs(deltaMass) < MinimumAminoacidSubstitutionDeltaMass)
{
continue;
}
result[ai].Add(new TargetVariant()
{
Source = ai.ToString(),
Target = new HashSet <string>(new[] { aj.ToString() }),
DeltaMass = deltaMass,
TargetType = VariantType.SingleAminoacidPolymorphism
});
}
}
foreach (var v in result.Values)
{
v.Sort((m1, m2) => m1.DeltaMass.CompareTo(m2.DeltaMass));
}
return(result);
}
public void TestIsMutationOneIgnoreNTerminal()
{
int site = -1;
Assert.IsTrue(MutationUtils.IsMutationOne2("ABCDEIR", "BBCDEIR", ref site, false, false, false));
Assert.IsFalse(MutationUtils.IsMutationOne2("ABCDEIR", "BBCDEIR", ref site, true, false, false));
Assert.IsTrue(MutationUtils.IsMutationOneIL2("ABCDEIR", "BBCDELR", ref site, false, false, false));
Assert.IsTrue(MutationUtils.IsMutationOneIL2("ABCDELR", "BBCDEIR", ref site, false, false, false));
Assert.IsFalse(MutationUtils.IsMutationOneIL2("ABCDELR", "BBCDEIR", ref site, true, false, false));
}
public void TestIsMutationOne()
{
//I -> L
Assert.IsTrue(MutationUtils.IsMutationOne("ABCDEIR", "ABCDELR"));
//no mutation
Assert.IsFalse(MutationUtils.IsMutationOne("ABCDEIR", "ABCDEIR"));
//two mutation
Assert.IsFalse(MutationUtils.IsMutationOne("ABCDEIR", "ABCDELK"));
int site = -1;
MutationUtils.IsMutationOne("ABCDEIR", "ABCDELR", ref site);
Assert.AreEqual(5, site);
}
private bool FindMutationOneType1(string pnovoseq, ref string source, ref int site)
{
if (!miss0group.ContainsKey(pnovoseq.Length))
{
return(false);
}
var lst = miss0group[pnovoseq.Length];
foreach (var m in lst)
{
if (MutationUtils.IsMutationOne2(m, pnovoseq, ref site, options.IgnoreNtermMutation, options.IgnoreDeamidatedMutation, options.IgnoreMultipleNucleotideMutation))
{
source = miss0[m];
//Console.WriteLine("TYPE1 : {0} ==> {1}", m, pnovoseq);
return(true);
}
}
return(false);
}
public void TestIsMutationOneIL()
{
int site = 0;
//C -> N
Assert.IsTrue(MutationUtils.IsMutationOneIL("ABCDEIR", "ABNDEIR", ref site));
Assert.AreEqual(2, site);
//C -> N, I -> L
Assert.IsTrue(MutationUtils.IsMutationOneIL("ABCDEIR", "ABNDELR", ref site));
Assert.AreEqual(2, site);
//I -> L
Assert.IsFalse(MutationUtils.IsMutationOneIL("ABCDEIR", "ABCDELR", ref site));
//no mutation
Assert.IsFalse(MutationUtils.IsMutationOneIL("ABCDEIR", "ABCDEIR", ref site));
//two mutation
Assert.IsFalse(MutationUtils.IsMutationOneIL("ABCDEIR", "ABCDEDK", ref site));
}
private static bool DoIsMutationOne(IsMutationOneDelegate func, string fromPeptide, string toPeptide, ref int mutationSite, bool ignoreNtermMutation, bool ignoreDeamidatedMutation, bool ignoreMultipleNucleotideMutation)
{
if (func(fromPeptide, toPeptide, ref mutationSite))
{
if (ignoreNtermMutation && 0 == mutationSite)
{
return(false);
}
if (ignoreDeamidatedMutation && MutationUtils.IsDeamidatedMutation(fromPeptide[mutationSite], toPeptide[mutationSite]))
{
return(false);
}
if (ignoreMultipleNucleotideMutation && IsSingleNucleotideMutation(fromPeptide[mutationSite], toPeptide[mutationSite]))
{
return(false);
}
return(true);
}
return(false);
}
private void GetCurrentImageFilename(out string mutationFilename, out string originalFilename)
{
if (gvPeptides.SelectedRows.Count == 0)
{
mutationFilename = string.Empty;
originalFilename = string.Empty;
return;
}
var spectrum = spectra[gvPeptides.SelectedRows[0].Index];
string imageFilename = GetImageFilename(spectrum);
var isMutation = (bool)(spectrum.Annotations["IsMutation"]);
var index = spectrum.Annotations["Index"].ToString();
var correspond = (from s in spectra
let cindex = s.Annotations["Index"].ToString()
let cIsMutation = (bool)(s.Annotations["IsMutation"])
where cindex.Equals(index) && s.Charge == spectrum.Charge && cIsMutation != isMutation
orderby s.Score descending
select s).ToList();
if (correspond.Count == 0)
{
string selectFilename = GetImageFilename(spectrum);
if (isMutation)
{
mutationFilename = selectFilename;
originalFilename = string.Empty;
}
else
{
mutationFilename = string.Empty;
originalFilename = selectFilename;
}
lblMutation.Text = string.Empty;
}
else
{
IIdentifiedSpectrum oriSp, mutSp;
if (isMutation)
{
mutSp = spectrum;
oriSp = correspond[0];
}
else
{
oriSp = spectrum;
mutSp = correspond[0];
}
mutationFilename = GetImageFilename(mutSp);
originalFilename = GetImageFilename(oriSp);
int mutationSite = -1;
var mutSeq = mutSp.Peptide.PureSequence;
var oriSeq = oriSp.Peptide.PureSequence;
MutationUtils.IsMutationOneIL(mutSeq, oriSeq, ref mutationSite);
if (mutationSite == 0)
{
lblMutation.Text = MyConvert.Format(".{0}.{1}\n.{2}.{3}",
oriSeq[mutationSite],
oriSeq.Substring(mutationSite + 1),
mutSeq[mutationSite],
mutSeq.Substring(mutationSite + 1));
}
else if (mutationSite == mutSeq.Length - 1)
{
lblMutation.Text = MyConvert.Format("{0}.{1}.\n{2}.{3}.",
oriSeq.Substring(0, mutationSite - 1),
oriSeq[mutationSite],
mutSeq.Substring(0, mutationSite - 1),
mutSeq[mutationSite]);
}
else
{
lblMutation.Text = MyConvert.Format("{0}.{1}.{2}\n{3}.{4}.{5}",
oriSeq.Substring(0, mutationSite - 1),
oriSeq[mutationSite],
oriSeq.Substring(mutationSite + 1),
mutSeq.Substring(0, mutationSite - 1),
mutSeq[mutationSite],
mutSeq.Substring(mutationSite + 1));
}
}
}
public void TestReplaceLToI()
{
Assert.AreEqual("A.DKDKDFLEI*DR.C", MutationUtils.ReplaceLToI("A.DKDKDFLEL*DR.C", "DKIKDFLEIDR"));
}
private string OutputPairedResult(Aminoacids aas, MascotPeptideTextFormat format, Dictionary <string, Sequence> proMap, IClassification <IIdentifiedPeptide> classification, string mutHeader, MascotPeptideTextFormat mutPepFormat, List <List <IGrouping <string, IIdentifiedPeptide> > > pairedMut, string pairedFile)
{
bool dbAnnotation = File.Exists(_uniprotXmlFile);
List <TempResult> tr = new List <TempResult>();
int resIndex = 1;
foreach (var res in pairedMut)
{
var mutCharges = (from r in res[0]
orderby r.Spectrum.Charge
select r.Spectrum.Charge).Distinct().ToList();
var pepCounts = GetPepCount(classification, res[0]);
int peplabel = 0;
bool bFound = false;
for (int pepIndex = 1; pepIndex < res.Count; pepIndex++)
{
var charges = (from r in res[pepIndex]
orderby r.Spectrum.Charge
select r.Spectrum.Charge).Distinct().ToList();
var commonCharges = mutCharges.Intersect(charges).ToList();
if (commonCharges.Count == 0)
{
continue;
}
if (!bFound)
{
bFound = true;
resIndex++;
}
peplabel++;
var curMutSpectrum = (from r in res[0]
where commonCharges.Contains(r.Spectrum.Charge)
orderby r.Spectrum.Score descending
select r).First();
var mutText = mutPepFormat.PeptideFormat.GetString(curMutSpectrum.Spectrum);
var curOriginalSpectrum = (from r in res[pepIndex]
where r.Spectrum.Charge == curMutSpectrum.Spectrum.Charge
orderby r.Spectrum.Score descending
select r).First();
var oriPureSeq = curOriginalSpectrum.PureSequence;
var mutFixSeq = MutationUtils.ReplaceLToI(curMutSpectrum.Sequence, oriPureSeq);
var mutFixPureSeq = PeptideUtils.GetPureSequence(mutFixSeq);
int mutationSite = -1;
string equalsToModification = string.Empty;
string rnaediting = string.Empty;
string databaseannotation = string.Empty;
bool isType1 = mutFixPureSeq.Length == oriPureSeq.Length;
if (isType1)
{
MutationUtils.IsMutationOneIL(mutFixPureSeq, oriPureSeq, ref mutationSite);
equalsToModification = spTable.GetModification(oriPureSeq[mutationSite], mutFixPureSeq[mutationSite]);
SnpCode.IsRnaEditing(aas[oriPureSeq[mutationSite]], aas[mutFixPureSeq[mutationSite]], out rnaediting);
}
else
{
mutationSite = Math.Min(mutFixPureSeq.Length, oriPureSeq.Length) - 1;
}
var pepMutation = MyConvert.Format("{0}{1}{2}", oriPureSeq[mutationSite], mutationSite + 1, mutFixPureSeq[mutationSite]);
List <Sequence> seqs = new List <Sequence>();
foreach (var p in curOriginalSpectrum.Proteins)
{
var ac = acParser.GetValue(p);
if (!proMap.ContainsKey(ac))
{
throw new Exception("Cannot find protein " + p + " in sequence database!");
}
seqs.Add(proMap[ac]);
}
var proMutations = (from p in curOriginalSpectrum.Proteins
let ac = acParser.GetValue(p)
let seq = proMap[ac]
let pos = seq.SeqString.IndexOf(oriPureSeq)
let pmu = MyConvert.Format("{0}{1}{2}", oriPureSeq[mutationSite], pos + mutationSite + 1, mutFixPureSeq[mutationSite])
select new { ProteinName = p, Mutation = pmu }).ToList();
var proMutation = (from pro in proMutations select pro.Mutation).Merge("/");
if (isType1 && dbAnnotation)
{
//sequence variants
foreach (var pro in proMutations)
{
var entry = GetUniprotEntry(pro.ProteinName);
if (entry == null)
{
continue;
}
foreach (var sv in entry.SequenceVariants)
{
var mut = string.Format("{0}{1}{2}", sv.Original, sv.Position, sv.Variation);
if (pro.Mutation.Equals(mut))
{
databaseannotation = string.Format("{0}=SequenceVariant {1}", pro.ProteinName, sv.Description);
break;
}
}
if (databaseannotation != string.Empty)
{
break;
}
}
//sequence conflicts
if (databaseannotation == string.Empty)
{
foreach (var pro in proMutations)
{
var entry = GetUniprotEntry(pro.ProteinName);
if (entry == null)
{
continue;
}
foreach (var sv in entry.SequenceConflicts)
{
if ((sv.BeginPosition != sv.EndPosition) || sv.Original.Length != 1)
{
continue;
}
var mut = string.Format("{0}{1}{2}", sv.Original, sv.BeginPosition, sv.Variation);
if (pro.Mutation.Equals(mut))
{
databaseannotation = string.Format("{0}=SequenceConflict {1}", pro.ProteinName, sv.Description);
break;
}
}
if (databaseannotation != string.Empty)
{
break;
}
}
}
}
List <string> proRefs = seqs.ConvertAll(m => m.Description).ToList();
int mutationCount;
var dnaMutation = aas[oriPureSeq[mutationSite]].TransferTo(aas[mutFixPureSeq[mutationSite]], out mutationCount);
var oriPepCounts = GetPepCount(classification, res[pepIndex]);
var line = string.Format("${0}-{1}\t{2}\t{3}\t{4}\t{5}\t{6}\t{7}\t{8}\t{9}\t{10}\t{11}\t{12}\t{13}\t{14}\t{15}\t{16}\t{17}",
resIndex,
peplabel,
mutText,
mutFixSeq,
mutFixPureSeq,
(from p in pepCounts
select p.ToString()).Merge("\t"),
pepMutation,
mutationCount,
dnaMutation,
curOriginalSpectrum.Spectrum.Query.FileScan.ShortFileName,
oriPureSeq,
curOriginalSpectrum.Proteins.Merge("/"),
proRefs.Merge("/"),
(from p in oriPepCounts
select p.ToString()).Merge("\t"),
proMutation,
equalsToModification,
rnaediting,
databaseannotation
);
tr.Add(new TempResult()
{
Line = line,
PepCount = pepCounts.Sum(),
OriginalSequence = oriPureSeq
});
}
var groups = tr.GroupBy(m => m.OriginalSequence).ToList();
groups.Sort((m1, m2) => - m1.Max(n => n.PepCount).CompareTo(m2.Max(n => n.PepCount)));
List <TempResult> lines = new List <TempResult>();
groups.ForEach(g => lines.AddRange(from l in g
orderby l.PepCount descending
select l));
using (StreamWriter sw = new StreamWriter(pairedFile))
{
string pepCountHeader = "";
string originalPepCountHeader = "";
foreach (var key in ClassificationSet.Keys)
{
pepCountHeader = pepCountHeader + "\t" + key + "_PepCount";
originalPepCountHeader = originalPepCountHeader + "\t" + key + "_OriginalCount";
}
sw.WriteLine("Index\t" + mutHeader + "\tSequence\tPureSequence" + pepCountHeader + "\tPepMutation\tDNAMutationCount\tDNAMutation\tOriginalFileScan\tOriginalSequence\tOriginalProteins\tOriginalReferences"
+ originalPepCountHeader + "\tProMutation\tEqualsToModification\tRNA-Editing\tDatabaseAnnotation");
lines.ForEach(m => sw.WriteLine(m.Line));
}
}
var pairedPeptideFile = pairedFile + ".peptides";
SavePeptidesFile(pairedMut, format, pairedPeptideFile);
return(pairedPeptideFile);
}
public override IEnumerable <string> Process(string fileName)
{
var aas = new Aminoacids();
Progress.SetMessage("reading pNovo result from " + pNovoPeptideFile + " ...");
var pNovoSpectra = new MascotPeptideTextFormat().ReadFromFile(pNovoPeptideFile);
var pNovoMap = new Dictionary <string, HashSet <string> >();
foreach (var pep in pNovoSpectra)
{
var key = pep.Query.FileScan.LongFileName;
if (!pNovoMap.ContainsKey(key))
{
pNovoMap[key] = new HashSet <string>();
}
pNovoMap[key].UnionWith(from p in pep.Peptides select p.PureSequence);
}
var format = new MascotPeptideTextFormat();
Progress.SetMessage("reading peptide-spectra-matches from " + fileName + " ...");
var spectra = format.ReadFromFile(fileName);
//价位筛选
spectra.RemoveAll(m => !charges.Contains(m.Charge));
//对于有不确定的氨基酸，直接忽略。
spectra.ForEach(m =>
{
for (int i = m.Peptides.Count - 1; i >= 0; i--)
{
if (m.Peptides[i].PureSequence.Any(n => aas[n].Codes.Length == 0))
{
m.RemovePeptideAt(i);
}
}
});
spectra.RemoveAll(m => m.Peptides.Count == 0);
Progress.SetMessage("comparing peptide-spectra-matches with pNovo result...");
//与pNovo判定的mutation是否一致？
spectra.RemoveAll(m =>
{
if (!IsMutationPeptide(m))
{
return(false);
}
var key = m.Query.FileScan.LongFileName;
if (!pNovoMap.ContainsKey(key))
{
return(true);
}
var set = pNovoMap[key];
return(!m.Peptides.Any(n => set.Contains(n.PureSequence.Replace('I', 'L'))));
});
//Get spectra whose peptides are all from mutated version
var mutSpectra = spectra.FindAll(m => IsMutationPeptide(m)).ToList();
var mutPeptides = (from s in mutSpectra
from p in s.Peptides
select p).ToList();
var mutGroup = mutPeptides.GroupBy(m => m.PureSequence);
//Get specra whose peptides are all from wide version
var fromSpectra = spectra.Except(mutSpectra).ToList();
fromSpectra.RemoveAll(m => m.Proteins.Any(n => mutationReg.Match(n).Success));
var fromPeptides = (from s in fromSpectra
from p in s.Peptides
select p).ToList();
var fromGroup = fromPeptides.GroupBy(m => m.PureSequence).ToGroupDictionary(n => n.Key.Length);
var minLength = fromGroup.Count == 0 ? 6 : fromGroup.Min(m => m.Key);
var maxLength = fromGroup.Count == 0 ? 30 : fromGroup.Max(m => m.Key);
//Check the mutation type
var type1 = new List <List <IGrouping <string, IIdentifiedPeptide> > >();
var type2 = new List <List <IGrouping <string, IIdentifiedPeptide> > >();
var type3 = new List <List <IGrouping <string, IIdentifiedPeptide> > >();
Progress.SetRange(0, mutGroup.Count());
Progress.SetPosition(0);
Progress.SetMessage("finding mutation-original pairs ...");
foreach (var mut in mutGroup)
{
var matched = new List <IGrouping <string, IIdentifiedPeptide> >();
matched.Add(mut);
Progress.Increment(1);
var protein = mut.First().Proteins[0];
List <List <IGrouping <string, IIdentifiedPeptide> > > type;
if (protein.EndsWith("type3"))
{
type = type3;
var mutseq = mut.Key.Substring(0, mut.Key.Length - 1);
for (int i = mut.Key.Length + 1; i <= maxLength; i++)
{
if (fromGroup.ContainsKey(i))
{
var others = fromGroup[i];
foreach (var o in others)
{
if (o.Key.StartsWith(mutseq))
{
matched.Add(o);
}
}
}
}
}
else if (protein.EndsWith("type2"))
{
type = type2;
for (int i = minLength; i < mut.Key.Length; i++)
{
if (fromGroup.ContainsKey(i))
{
var others = fromGroup[i];
foreach (var o in others)
{
var oseq = o.Key.Substring(0, o.Key.Length - 1);
if (mut.Key.StartsWith(oseq))
{
matched.Add(o);
}
}
}
}
}
else if (protein.EndsWith("type1"))
{
type = type1;
if (fromGroup.ContainsKey(mut.Key.Length))
{
var oLength = fromGroup[mut.Key.Length];
foreach (var o in oLength)
{
int mutationSite = -1;
if (MutationUtils.IsMutationOneIL2(o.Key, mut.Key, ref mutationSite, IgnoreNtermMutation, IgnoreDeamidatedMutation, IgnoreMultipleNucleotideMutation))
{
matched.Add(o);
}
}
}
}
else
{
throw new Exception("There is no mutation type information at protein name: " + protein + "\nIt should be like MUL_NHLGQK_type1, MUL_NHLGQK_type2 or MUL_NHLGQK_type3");
}
type.Add(matched);
}
type1.Sort((m1, m2) =>
{
var res = m1.Count.CompareTo(m2.Count);
if (res == 0)
{
res = m2[0].Count().CompareTo(m1[0].Count());
}
return(res);
});
Progress.SetMessage("reading protein sequences ...");
var proteins = SequenceUtils.Read(new FastaFormat(), fastaFile);
var proMap = proteins.ToDictionary(m =>
{
string ac;
if (acParser.TryParse(m.Name, out ac))
{
return(ac);
}
else
{
return(m.Name);
}
});
var classification = GetClassification();
string mutHeader = "FileScan\tMH+\tDiff(MH+)\tCharge\tRank\tScore\tExpectValue\tModification";
var mutPepFormat = new MascotPeptideTextFormat(mutHeader);
Progress.SetMessage("writing result ...");
var result1 = DoStatistic(fileName, aas, format, proMap, classification, mutHeader, mutPepFormat, type1, ".type1");
var result2 = DoStatistic(fileName, aas, format, proMap, classification, mutHeader, mutPepFormat, type2, ".type2");
var result3 = DoStatistic(fileName, aas, format, proMap, classification, mutHeader, mutPepFormat, type3, ".type3");
return(result1.Concat(result2).Concat(result3).ToArray());
}
