public override IEnumerable <string> Process(string useless)
{
var result = new MappedMirnaGroupXmlFileFormat().ReadFromFile(options.InputFile);
using (StreamWriter sw = new StreamWriter(options.OutputFile))
{
sw.WriteLine("miRNA\tLocation\tTotalCount\tPerfectMatch\tMiss5_2\tMiss3_3\tMissInternal");
foreach (var res in result)
{
var items = res.GetAlignedLocations();
if (res.DisplayName.Equals("hsa-mir-486-5p:TCCTGTACTGAGCTGCCCCGAG"))
{
items.ForEach(m => Console.WriteLine(m.Parent.Qname + "\t" + m.Strand + "\t" + m.MismatchPositions));
}
var pmcount = items.Count(m => m.NumberOfMismatch == 0);
var mis5 = items.Count(m =>
{
SAMAlignedLocation loc = m;
if (loc.NumberOfMismatch == 0)
{
return(false);
}
var mp = loc.MismatchPositions;
if (loc.Strand == '-')
{
mp = new string(mp.Reverse().ToArray());
}
return(reg5.Match(mp).Success);
});
var mis3 = items.Count(m =>
{
var loc = m;
if (loc.NumberOfMismatch == 0)
{
return(false);
}
var mp = loc.MismatchPositions;
if (loc.Strand == '+')
{
mp = new string(mp.Reverse().ToArray());
}
return(reg3.Match(mp).Success);
});
sw.WriteLine("{0}\t{1}\t{2}\t{3}\t{4}\t{5}\t{6}", res.DisplayName, res.DisplayLocation, items.Count, pmcount, mis5, mis3, items.Count - pmcount - mis5 - mis3);
}
}
return(new string[] { options.OutputFile });
}
public override IEnumerable <string> Process()
{
var countfiles = options.GetCountFiles();
var dic = new Dictionary <string, Dictionary <string, MappedMirnaGroup> >();
foreach (var file in countfiles)
{
Progress.SetMessage("Reading miRNA mapped file " + file.File + " ...");
var mirnas = new MappedMirnaGroupXmlFileFormat().ReadFromFile(file.File);
dic[file.Name] = mirnas.ToDictionary(m => m.DisplayName);
}
var features = (from c in dic.Values
from k in c.Keys
select k).Distinct().OrderBy(m => m).ToList();
var names = dic.Keys.OrderBy(m => m).ToList();
using (StreamWriter sw = new StreamWriter(options.OutputFile))
using (StreamWriter swNTA = new StreamWriter(options.NTAFile))
using (StreamWriter swIso = new StreamWriter(options.IsomirFile))
using (StreamWriter swIsoNTA = new StreamWriter(options.IsomirNTAFile))
{
sw.WriteLine("Feature\tLocation\tSequence\t{0}", names.Merge("\t"));
swNTA.WriteLine("Feature\tLocation\tSequence\t{0}", names.Merge("\t"));
swIso.WriteLine("Feature\tLocation\tSequence\t{0}", names.Merge("\t"));
swIsoNTA.WriteLine("Feature\tLocation\tSequence\t{0}", names.Merge("\t"));
foreach (var feature in features)
{
OutputCount(sw, dic, feature, names, MirnaConsts.NO_OFFSET, false, "");
OutputCount(swNTA, dic, feature, names, MirnaConsts.NO_OFFSET, true, "");
OutputCount(swIso, dic, feature, names, 0, false, "_+_0");
OutputCount(swIso, dic, feature, names, 1, false, "_+_1");
OutputCount(swIso, dic, feature, names, 2, false, "_+_2");
OutputCount(swIsoNTA, dic, feature, names, 0, true, "_+_0");
OutputCount(swIsoNTA, dic, feature, names, 1, true, "_+_1");
OutputCount(swIsoNTA, dic, feature, names, 2, true, "_+_2");
}
}
var result = new[] { options.OutputFile, options.IsomirFile, options.NTAFile, options.IsomirNTAFile }.ToList();
return(result);
}
public List <MappedMirnaGroup> Build(string countXmlFile)
{
var result = new MappedMirnaGroupXmlFileFormat().ReadFromFile(countXmlFile);
foreach (var group in result)
{
foreach (var mirna in group)
{
foreach (var region in mirna.MappedRegions)
{
var positions = region.Mapped.Keys.ToList();
foreach (var position in positions)
{
var mapped = region.Mapped[position];
mapped.AlignedLocations.RemoveAll(q =>
{
var snp = q.GetNotGsnapMismatch(q.Parent.Sequence);
if (null == snp)
{
return(true);
}
return(!snp.IsMutation('T', 'C'));
});
if (mapped.AlignedLocations.Count == 0)
{
region.Mapped.Remove(position);
}
}
}
mirna.MappedRegions.RemoveAll(l => l.Mapped.Count == 0);
}
group.RemoveAll(n => n.MappedRegions.Count == 0);
}
result.RemoveAll(m => m.Count == 0);
return(result);
}
public override IEnumerable <string> Process()
{
var format = new MappedMirnaGroupXmlFileFormat();
Progress.SetMessage("reading mapped reads from " + options.ReferenceFile + " ...");
var refitems = format.ReadFromFile(options.ReferenceFile);
var refSpecies = refitems[0][0].Name.StringBefore("-");
Progress.SetMessage("reading mapped reads from " + options.SampleFile + " ...");
var samitems = format.ReadFromFile(options.SampleFile);
var samSpecies = samitems[0][0].Name.StringBefore("-");
var paired = GetPairedMiRNA(refitems, samitems);
//using (StreamWriter sw = new StreamWriter(targetFile))
//{
// sw.WriteLine("microRNA\t{0}_sequence\t{1}_sequence\tbp_difference\tquery_sequence\t{0}_count\t{0}_estimate_count\t{1}_count\t{1}_estimate_count", refName, samName);
// var keys = refitems.Keys.Union(samitems.Keys).Distinct().OrderBy(m => m).ToList();
// foreach (var key in keys)
// {
// sw.Write(key);
// string refseq = string.Empty;
// string samseq = string.Empty;
// MappedMiRNA refmirna = null;
// MappedMiRNA sammirna = null;
// Dictionary<string, List<QueryMapped>> refmirnamap = new Dictionary<string, List<QueryMapped>>();
// Dictionary<string, List<QueryMapped>> sammirnamap = new Dictionary<string, List<QueryMapped>>();
// if (refitems.ContainsKey(key))
// {
// refmirna = refitems[key];
// refseq = refmirna.Name.Contains(":") ? refmirna.Name.StringAfter(":") : string.Empty;
// refmirnamap = ConvertToMap(refmirna);
// }
// if (samitems.ContainsKey(key))
// {
// sammirna = samitems[key];
// samseq = sammirna.Name.Contains(":") ? sammirna.Name.StringAfter(":") : string.Empty;
// sammirnamap = ConvertToMap(sammirna);
// }
// var seqs = refmirnamap.Keys.Union(sammirnamap.Keys).OrderBy(m => m).ToList();
// CombinedSequence cs = null;
// if (!string.IsNullOrEmpty(refseq) && !string.IsNullOrEmpty(samseq))
// {
// cs = MirnaUtils.GetCombinedSequence(refseq, samseq);
// sw.Write("\t{0}\t{1}\t{2}", cs.GetAnnotatedSequence1(), cs.GetAnnotatedSequence2(), cs.MismatchPositions.Length);
// }
// else
// {
// sw.Write("\t{0}\t{1}\t-", refseq, samseq);
// }
// sw.WriteLine();
// foreach (var seq in seqs)
// {
// sw.Write("\t\t\t\t{0}", seq);
// if (refmirnamap.ContainsKey(seq))
// {
// sw.Write("\t{0:0.00}", refmirnamap[seq].Sum(m => m.QueryCount));
// sw.Write("\t{0:0.00}", refmirnamap[seq].Sum(m => m.EsminatedCount));
// }
// else
// {
// sw.Write("\t\t");
// }
// if (sammirnamap.ContainsKey(seq))
// {
// sw.Write("\t{0:0.00}", sammirnamap[seq].Sum(m => m.QueryCount));
// sw.Write("\t{0:0.00}", sammirnamap[seq].Sum(m => m.EsminatedCount));
// }
// else
// {
// sw.Write("\t\t");
// }
// sw.WriteLine();
// }
// }
//}
return(new string[] { options.OutputFile });
}