public bool Accept(IIdentifiedSpectrum t)
{
foreach (var p in t.Proteins)
{
var ac = acParser.GetValue(p);
return(conMap.Contains(ac));
}
return(false);
}
public bool Accept(IIdentifiedProteinGroup t)
{
foreach (var p in t)
{
var ac = acParser.GetValue(p.Name);
if (conMap.Contains(ac))
{
return(true);
}
}
return(false);
}
/// <summary>
/// 用蛋白质的AccessNumber取代原来的蛋白名，以避免SEQUEST搜索中产生的相同蛋白质有长短不一样的名字导致的错误
/// </summary>
/// <typeparam name="T">鉴定谱图</typeparam>
/// <param name="spectra">鉴定谱图列表</param>
/// <param name="proteinAccessNumberParser">蛋白质AccessNumber解析器</param>
public static void RefineIdentifiedProteinName <T>(List <T> spectra, IStringParser <string> proteinAccessNumberParser) where T : IIdentifiedSpectrum
{
foreach (T spectrum in spectra)
{
foreach (IIdentifiedPeptide peptide in spectrum.Peptides)
{
ReadOnlyCollection <string> proteinNames = peptide.Proteins;
for (int i = 0; i < proteinNames.Count; i++)
{
peptide.SetProtein(i, proteinAccessNumberParser.GetValue(proteinNames[i]));
}
}
}
}
public static Dictionary<string, Sequence> GetAccessNumberMap(string database, IStringParser<string> acParser)
{
Dictionary<string, Sequence> result = new Dictionary<string, Sequence>();
using (StreamReader sr = new StreamReader(database))
{
FastaFormat sf = new FastaFormat();
Sequence seq;
while ((seq = sf.ReadSequence(sr)) != null)
{
string ac = acParser.GetValue(seq.Name);
result[ac] = seq;
}
}
return result;
}
public static Dictionary <string, Sequence> GetAccessNumberMap(string database, IStringParser <string> acParser)
{
Dictionary <string, Sequence> result = new Dictionary <string, Sequence>();
using (StreamReader sr = new StreamReader(database))
{
FastaFormat sf = new FastaFormat();
Sequence seq;
while ((seq = sf.ReadSequence(sr)) != null)
{
string ac = acParser.GetValue(seq.Name);
result[ac] = seq;
}
}
return(result);
}
public static HashSet <string> GetContaminationAccessNumbers(IStringParser <string> acParser, string fastaFilename, string contaminationDescriptionPattern,
IProgressCallback progress)
{
HashSet <string> result = new HashSet <string>();
if (progress == null)
{
progress = new EmptyProgressCallback();
}
Regex reg = new Regex(contaminationDescriptionPattern, RegexOptions.IgnoreCase);
progress.SetMessage("Get contamination map from database ...");
var ff = new FastaFormat();
using (var sr = new StreamReader(fastaFilename))
{
progress.SetRange(1, sr.BaseStream.Length);
Sequence seq;
while ((seq = ff.ReadSequence(sr)) != null)
{
if (progress.IsCancellationPending())
{
throw new UserTerminatedException();
}
progress.SetPosition(sr.GetCharpos());
string ac = acParser.GetValue(seq.Name);
if (reg.Match(seq.Reference).Success)
{
result.Add(ac);
}
}
}
progress.SetMessage("Get contamination map from database finished.");
return(result);
}
protected override bool FillSequence(IIdentifiedResult result)
{
foreach (var msfFile in msfFiles)
{
var proteins = new MsfDatabaseToNoredundantProcessor().ParseProteinSequences(msfFile).ToDictionary(m => acParser.GetValue(m.Name));
foreach (var g in result)
{
foreach (var p in g)
{
var name = acParser.GetValue(p.Name);
Sequence seq;
if (proteins.TryGetValue(name, out seq))
{
p.Sequence = seq.SeqString;
p.Reference = seq.Reference;
p.Name = name;
}
}
}
}
foreach (var g in result)
{
foreach (IIdentifiedProtein protein in g)
{
if (protein.Name.StartsWith(MsfDatabaseParser.DECOY_PREFIX))
{
continue;
}
if (protein.Sequence == null)
{
throw new Exception(
MyConvert.Format(
"Couldn't find sequence of protein {0}, change access number pattern or select another database.",
protein.Name));
}
}
}
return(true);
}
public static void FillSequenceFromFasta(IStringParser <string> acParser, string fastaFilename, IIdentifiedResult t,
IProgressCallback progress)
{
if (progress == null)
{
progress = new EmptyProgressCallback();
}
progress.SetMessage("Initializing accessNumber/protein map ...");
var acMap = new Dictionary <string, IIdentifiedProtein>();
foreach (IIdentifiedProteinGroup group in t)
{
foreach (IIdentifiedProtein protein in group)
{
string ac = acParser.GetValue(protein.Name);
if (acMap.ContainsKey(ac))
{
throw new Exception("Duplicate access number " + ac);
}
acMap[ac] = protein;
if (ac != protein.Name)
{
if (acMap.ContainsKey(protein.Name))
{
throw new Exception("Duplicate access number " + protein.Name);
}
acMap[protein.Name] = protein;
}
}
}
progress.SetMessage("Filling sequence from database ...");
var ff = new FastaFormat();
using (var sr = new StreamReader(fastaFilename))
{
progress.SetRange(1, sr.BaseStream.Length);
Sequence seq;
while ((seq = ff.ReadSequence(sr)) != null)
{
if (progress.IsCancellationPending())
{
throw new UserTerminatedException();
}
progress.SetPosition(sr.BaseStream.Position);
string ac = acParser.GetValue(seq.Name);
if (acMap.ContainsKey(ac))
{
IIdentifiedProtein protein = acMap[ac];
protein.Name = seq.Name.Replace("/", " ");
protein.Description = seq.Description.Replace("\t", " ").Replace("/", " ");
protein.Sequence = seq.SeqString;
}
}
}
var failed = acMap.Values.Where(l => l.Sequence == null).ToList();
if (failed.Count > 0)
{
var proteinNames = failed.ConvertAll(l => l.Name).ToArray();
if (!proteinNames.All(l => l.StartsWith("XXX_")))
{
throw new Exception(string.Format("Couldn't find sequence of following protein(s), change access number pattern or select another database\n{0}", proteinNames.Merge("/")));
}
}
progress.SetMessage("Fill sequence from database finished.");
}
