private void WriteInfoFile(List <string> result, string resultFilename, ReadSummary readSummary, List <FeatureItemGroup> allmapped)
{
var infoFile = Path.ChangeExtension(resultFilename, ".info");
WriteSummaryFile(infoFile, readSummary, allmapped);
result.Add(infoFile);
}
protected ReadSummary GetReadSummary(List <FeatureItemGroup> allmapped, HashSet <string> excludeQueries, List <SAMAlignedItem> reads, List <QueryInfo> totalQueries)
{
var result = new ReadSummary();
if (File.Exists(options.CountFile))
{
result.TotalRead = Counts.GetTotalCount();
}
else
{
result.TotalRead = totalQueries.Count;
}
var featureQueries = new HashSet <string>(from fig in allmapped
from fi in fig
from loc in fi.Locations
from sl in loc.SamLocations
select sl.SamLocation.Parent.OriginalQname);
result.FeatureRead = featureQueries.Sum(l => Counts.GetCount(l));
result.ExcludeRead = excludeQueries.Sum(l => Counts.GetCount(l));
result.GenomeRead = (from query in totalQueries
where (!query.Name.Contains(SmallRNAConsts.NTA_TAG) || query.Name.EndsWith(SmallRNAConsts.NTA_TAG))
let originalQname = query.Name.StringBefore(SmallRNAConsts.NTA_TAG)
where !featureQueries.Contains(originalQname) && query.Mismatch == 0 && query.Length >= options.TooShortReadLength
select originalQname).Distinct().Sum(m => Counts.GetCount(m));
if (Counts.ItemMap != null)
{
result.TooShortRead = (from read in Counts.ItemMap.Values
where !featureQueries.Contains(read.Qname) && read.SequenceLength < 20
select read.Count).Sum();
}
else
{
result.TooShortRead = 0;
}
return(result);
}
protected void WriteSummaryFile(string infoFile, ReadSummary readSummary, List <FeatureItemGroup> allmapped)
{
if (!File.Exists(infoFile) || !options.NotOverwrite)
{
Progress.SetMessage("summarizing ...");
using (var sw = new StreamWriter(infoFile))
{
WriteOptions(sw);
if (File.Exists(options.CountFile))
{
sw.WriteLine("#countFile\t{0}", options.CountFile);
}
sw.WriteLine("TotalReads\t{0}", readSummary.TotalRead);
if (readSummary.ExcludeRead > 0)
{
sw.WriteLine("ExcludedReads\t{0}", readSummary.ExcludeRead);
}
sw.WriteLine("MappedReads\t{0}", readSummary.MappedRead);
sw.WriteLine("FeatureReads\t{0}", readSummary.FeatureRead);
sw.WriteLine("GenomeReads\t{0}", readSummary.GenomeRead);
sw.WriteLine("TooShortReads\t{0}", readSummary.TooShortRead);
sw.WriteLine("UnannotatedReads\t{0}", readSummary.UnannotatedRead);
//Output individual category
foreach (var cat in SmallRNAConsts.Biotypes)
{
var count = (from c in allmapped
from l in c
where l.Name.StartsWith(cat)
select l.GetEstimatedCount()).Sum();
if (count > 0)
{
sw.WriteLine("{0}\t{1:0.#}", cat, count);
}
}
}
}
}