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 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 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());
}
