public override IEnumerable <string> Process()
{
var format = new MascotPeptideTextFormat();
Progress.SetMessage("reading peptide-spectra-matches from " + options.PeptideFile + " ...");
var spectra = format.ReadFromFile(options.PeptideFile);
var seqMap = new Dictionary <string, IIdentifiedPeptide>();
foreach (var spec in spectra)
{
seqMap[spec.Peptide.PureSequence] = spec.Peptide;
}
var aas = (from c in new Aminoacids().GetVisibleAminoacids()
where c != 'I'
select c.ToString()).Merge("");
var ff = new FastaFormat();
Progress.SetMessage("inserting amino acid ...");
using (var sw = new StreamWriter(options.OutputFile))
{
sw.WriteLine(File.ReadAllText(options.DatabaseFile));
var seqs = seqMap.Keys.OrderBy(m => m).ToArray();
var reversed_index = 1000000;
foreach (var seq in seqs)
{
for (int i = 0; i < seq.Length; i++)
{
for (int j = 0; j < aas.Length; j++)
{
var newsequence = seq.Insert(i, aas[j].ToString());
var newref = string.Format("INS_{0}_{1}{2} Insertion of {3}", seq, i, aas[j], seqMap[seq].Proteins.Merge("/"));
var newseq = new Sequence(newref, newsequence);
ff.WriteSequence(sw, newseq);
if (options.GenerateReversedPeptide)
{
var revsequence = SequenceUtils.GetReversedSequence(newsequence);
var revref = string.Format("REVERSED_{0}", reversed_index++);
var revseq = new Sequence(revref, revsequence);
ff.WriteSequence(sw, revseq);
}
}
}
}
}
return(new[] { options.OutputFile });
}
public static void WriteFastaFile(string fastaFilename, IList <IIdentifiedProteinGroup> t, Func <IIdentifiedProteinGroup, bool> validateGroup)
{
foreach (var g in t)
{
if (validateGroup(g) && g.Count > 0 && (g[0].Sequence == null || g[0].Sequence.Length == 0))
{
return;
}
}
var ff = new FastaFormat();
using (var sw = new StreamWriter(fastaFilename))
{
foreach (IIdentifiedProteinGroup mpg in t)
{
if (validateGroup(mpg))
{
foreach (IIdentifiedProtein protein in mpg)
{
ff.WriteSequence(sw, protein.Reference, protein.Sequence);
}
}
}
}
}
public override IEnumerable <string> Process(string fileName)
{
DatFormat reader = new DatFormat();
FastaFormat writer = new FastaFormat();
string result = FileUtils.ChangeExtension(fileName, ".fasta");
long fileLength = new FileInfo(fileName).Length;
using (StreamReader sr = new StreamReader(fileName))
using (StreamWriter sw = new StreamWriter(result))
{
Progress.SetRange(0, fileLength);
Sequence seq;
while ((seq = reader.ReadSequence(sr)) != null)
{
if (Progress.IsCancellationPending())
{
throw new UserTerminatedException();
}
Progress.SetPosition(sr.GetCharpos());
writer.WriteSequence(sw, seq);
}
}
return(new string[] { result });
}
public override IEnumerable <string> Process(string fileName)
{
string result = FileUtils.ChangeExtension(fileName, "") + "_" + name + new FileInfo(fileName).Extension;
FastaFormat format = new FastaFormat();
Progress.SetMessage("Processing " + fileName);
using (StreamReader sr = new StreamReader(fileName))
{
Progress.SetRange(0, sr.BaseStream.Length);
using (StreamWriter sw = new StreamWriter(result))
{
Sequence seq;
while ((seq = format.ReadSequence(sr)) != null)
{
if (Progress.IsCancellationPending())
{
throw new UserTerminatedException();
}
Progress.SetPosition(sr.BaseStream.Position);
if (nameRegex.Match(seq.Name).Success)
{
format.WriteSequence(sw, seq);
}
}
}
}
return(new string[] { result });
}
private void WriteFasta(StreamWriter swFasta, Dictionary <string, Sequence> seqMap, string protein)
{
if (swFasta != null)
{
string[] parts = Regex.Split(protein, @"\s+");
ff.WriteSequence(swFasta, seqMap[parts[1].Trim()]);
}
}
private void ProcessFile(ref int index, StreamWriter sw, string fastaFile, bool isContaminant)
{
FastaFormat ff = new FastaFormat();
using (StreamReader sr = new StreamReader(fastaFile))
{
Progress.SetRange(0, sr.BaseStream.Length);
Sequence seq;
while ((seq = ff.ReadSequence(sr)) != null)
{
Progress.SetPosition(sr.BaseStream.Position);
if (isContaminant)
{
if (!seq.Reference.StartsWith("CON_"))
{
seq.Reference = "CON_" + seq.Reference;
}
}
if (!options.ReversedOnly)
{
ff.WriteSequence(sw, seq);
}
if (options.IsPseudoAminoacid)
{
options.PseudoAminoacidBuilder.Build(seq);
}
index++;
Sequence reversedSeq = GetReversedSequence(index, seq);
ff.WriteSequence(sw, reversedSeq);
}
}
}
public override IEnumerable <string> Process(string fileName)
{
var parser = new MsfDatabaseParser(SearchEngineType.SEQUEST);
var seqs = parser.ParseProteinSequences(fileName);
SQLiteDBHelper sqlite = new SQLiteDBHelper(fileName);
var result = new List <Sequence>();
var aaReader = sqlite.ExecuteReader("select count(*) from peptides_decoy", null);
if (aaReader.Read())
{
if (aaReader.GetInt32(0) > 0) // there are decoy database
{
foreach (var seq in seqs)
{
result.Add(seq);
var revseq = new Sequence(MsfDatabaseParser.GetReversedReference(seq.Reference), SequenceUtils.GetReversedSequence(seq.SeqString));
result.Add(revseq);
}
}
}
if (result.Count == 0)
{
result = seqs;
}
var fastafile = fileName + ".fasta";
using (var sw = new StreamWriter(fastafile))
{
var ff = new FastaFormat();
foreach (var seq in result)
{
ff.WriteSequence(sw, seq);
}
}
return(new[] { fastafile });
}
public override IEnumerable <string> Process(string fileName)
{
FastaFormat ff = new FastaFormat();
var result = Path.ChangeExtension(fileName, ".dM.fasta");
using (StreamReader sr = new StreamReader(fileName))
using (StreamWriter sw = new StreamWriter(result))
{
Sequence seq;
Progress.SetRange(1, sr.BaseStream.Length);
while ((seq = ff.ReadSequence(sr)) != null)
{
Progress.SetPosition(StreamUtils.GetCharpos(sr));
if (seq.SeqString.StartsWith("M"))
{
seq.SeqString = seq.SeqString.Substring(1);
seq.Reference = seq.Name + " N-terminal-M-Removed " + seq.Description;
}
ff.WriteSequence(sw, seq);
}
}
return(new string[] { result });
}
public override IEnumerable <string> Process(string fileName)
{
var result = new List <string>();
string[] acLines = File.ReadAllLines(fileName);
var acs = new HashSet <string>();
foreach (var acline in acLines)
{
string ac;
if (!parser.TryParse(acline, out ac))
{
ac = acline;
}
acs.Add(ac);
}
var findAcs = new HashSet <string>();
var resultFile = fileName + ".fasta";
result.Add(resultFile);
var ff = new FastaFormat();
using (StreamWriter sw = new StreamWriter(resultFile))
using (StreamReader sr = new StreamReader(database))
{
Progress.SetRange(0, sr.BaseStream.Length);
Sequence seq;
while ((seq = ff.ReadSequence(sr)) != null)
{
Progress.SetPosition(sr.BaseStream.Position);
string curAc;
if (!parser.TryParse(seq.Name, out curAc))
{
curAc = seq.Name;
}
if (acs.Contains(curAc))
{
findAcs.Add(curAc);
if (this.replaceName)
{
seq.Reference = curAc;
}
ff.WriteSequence(sw, seq);
}
}
}
acs.ExceptWith(findAcs);
var missFile = fileName + ".miss";
if (acs.Count > 0)
{
using (StreamWriter sw = new StreamWriter(missFile))
{
foreach (var ac in acs)
{
sw.WriteLine(ac);
}
}
result.Add(missFile);
}
else if (File.Exists(missFile))
{
File.Delete(missFile);
}
return(result);
}
public override IEnumerable <string> Process()
{
var srItems = SequenceRegionUtils.GetSequenceRegions(options.InputFile).Where(m => options.AcceptName(m.Name)).ToList();
srItems = (from sr in srItems.GroupBy(m => m.Name)
select sr.First()).ToList();
var keepChrInName = options.KeepChrInName && srItems.Any(m => m.Name.StartsWith("chr"));
if (!keepChrInName)
{
srItems.ForEach(m => m.Seqname = m.Seqname.StringAfter("chr"));
}
var srMap = srItems.ToGroupDictionary(m => m.Seqname);
var ff = new FastaFormat(int.MaxValue);
using (StreamWriter sw = new StreamWriter(options.OutputFile))
{
using (StreamReader sr = new StreamReader(options.GenomeFastaFile))
{
Sequence seq;
while ((seq = ff.ReadSequence(sr)) != null)
{
Progress.SetMessage("processing " + seq.Name + " ...");
var name = seq.Name;
if (!keepChrInName)
{
name = name.StringAfter("chr");
}
List <GtfItem> items;
if (!srMap.TryGetValue(name, out items))
{
if (name.Equals("M"))
{
name = "MT";
srMap.TryGetValue(name, out items);
}
else if (name.Equals("chrM"))
{
name = "chrMT";
srMap.TryGetValue(name, out items);
}
else if (name.Equals("MT"))
{
name = "M";
srMap.TryGetValue(name, out items);
}
else if (name.Equals("chrMT"))
{
name = "chrM";
srMap.TryGetValue(name, out items);
}
}
if (items != null)
{
Progress.SetMessage(" there are {0} entries in {1} ...", items.Count, name);
foreach (var item in items)
{
if (item.Start - 1 + item.Length >= seq.SeqString.Length)
{
throw new Exception(string.Format("{0} exceed chromosome {1} length {2}", item, name, seq.SeqString.Length));
}
var newseq = seq.SeqString.Substring((int)item.Start - 1, (int)item.Length);
if (item.Strand == '-')
{
newseq = SequenceUtils.GetReverseComplementedSequence(newseq);
}
newseq = newseq.ToUpper();
var newname = string.Format("{0} {1} {2}", item.Name, item.GetLocationWithoutStrand(), item.Strand);
var entry = new Sequence(newname, newseq);
ff.WriteSequence(sw, entry);
}
}
}
}
}
return(new string[] { options.OutputFile });
}
