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);
}
/// <summary>
/// 从isobaricFile中读取spectra对应的isobaric labelling信息。
/// </summary>
/// <param name="spectra"></param>
/// <param name="isobaricFile"></param>
/// <param name="progress"></param>
public static void Load(List <IIdentifiedSpectrum> spectra, string isobaricFile, bool readPeaks = false, IProgressCallback progress = null)
{
if (progress == null)
{
progress = new EmptyProgressCallback();
}
var fileNames = new HashSet <string>(from s in spectra
let fs = s.Query.FileScan
select fs.Experimental + "," + fs.FirstScan.ToString());
using (var reader = IsobaricResultFileFormatFactory.GetXmlReader(true, readPeaks))
{
var usedChannels = IsobaricScanXmlUtils.GetUsedChannels(isobaricFile);
reader.Open(isobaricFile);
progress.SetMessage("Reading Isobaric from {0} ...", isobaricFile);
progress.SetRange(1, spectra.Count);
foreach (var spectrum in spectra)
{
if (progress.IsCancellationPending())
{
throw new UserTerminatedException();
}
progress.Increment(1);
var fs = spectrum.Query.FileScan;
if (reader.Has(fs.Experimental, fs.FirstScan))
{
spectrum.SetIsobaricItem(reader.Read(fs.Experimental, fs.FirstScan, usedChannels));
}
else
{
spectrum.SetIsobaricItem(null);
}
}
}
}
/// <summary>
/// Fill dbsnp information. The name of SNPItem will be replaced by dbSNP name and the mapping between dbSNP name and old SNPItem name will be returned.
/// </summary>
/// <param name="snpItems"></param>
/// <param name="dbSnpVcfFile"></param>
/// <param name="progress"></param>
/// <returns></returns>
public static Dictionary <string, string> FillDbsnpIdByPosition(this IEnumerable <SNPItem> snpItems, string dbSnpVcfFile, IProgressCallback progress = null)
{
var sourceDbsnpMap = snpItems.ToDictionary(m => m.Name, m => m.Name);
if (progress == null)
{
progress = new EmptyProgressCallback();
}
var dic = snpItems.ToDoubleDictionary(m => m.Chrom, m => m.Position);
progress.SetMessage("Filling dbSNP id from {0} ...", dbSnpVcfFile);
using (var sr = new StreamReader(dbSnpVcfFile))
{
progress.SetRange(0, sr.BaseStream.Length);
string line;
while ((line = sr.ReadLine()) != null)
{
if (!line.StartsWith("##"))
{
break;
}
}
int linecount = 0;
Dictionary <int, SNPItem> chrMap = null;
int lastChr = -1;
while (line != null)
{
linecount++;
if (linecount % 10000 == 0)
{
progress.SetPosition(sr.GetCharpos());
}
try
{
//make sure it is SNV
if (!line.Contains("VC=SNV"))
{
continue;
}
//Even it marked as SNV, it still could be insertion/deletion
//2 179658175 rs11537855 C CC,CT . . RS=11537855;RSPOS=179658175;dbSNPBuildID=120;SSR=0;SAO=0;VP=0x050100001205000002000110;GENEINFO=TTN:7273;WGT=1;VC=SNV;SLO;NSF;REF;ASP;OTHERKG;NOC
var parts = line.Split('\t');
if (parts[3].Split(',').Any(l => l.Length != 1))
{
continue;
}
if (parts[4].Split(',').Any(l => l.Length != 1))
{
continue;
}
var chr = HumanChromosomeToInt(parts[0]);
var position = int.Parse(parts[1]);
if (lastChr != chr)
{
if (!dic.TryGetValue(chr, out chrMap))
{
continue;
}
lastChr = chr;
}
SNPItem source;
if (!chrMap.TryGetValue(position, out source))
{
continue;
}
if (!source.Name.Equals(parts[2]))
{
sourceDbsnpMap.Remove(source.Name);
sourceDbsnpMap[source.Name] = parts[2];
}
source.DbsnpRefAllele = parts[3][0];
source.DbsnpAltAllele = parts[4][0];
source.DbsnpIsReversed = parts[7].Contains(";RV;");
}
finally
{
line = sr.ReadLine();
}
}
}
var snpMap = snpItems.ToDictionary(m => m.Name);
var result = new Dictionary <string, string>();
foreach (var r in sourceDbsnpMap)
{
result[r.Value] = r.Key;
if (!r.Key.Equals(r.Value))
{
snpMap[r.Key].Name = r.Value;
}
}
progress.SetMessage("Filling dbSNP id finished.");
return(result);
}
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.");
}
