public void DrawPositionImage(List <SAMAlignedItem> notNTAreads, List <FeatureLocation> features, string category, string positionFile, bool positionByPercentage = false, int minimumReadCount = 5)
{
if (features.Count > 0)
{
Progress.SetMessage("mapping reads overlapped to {0} {1} entries.", features.Count, category);
//map reads to miRNA without considering NTA and offset.
var map = BuildStrandMap(notNTAreads);
new SmallRNAMapper(category, options, m => true).MapReadToFeature(features, map);
SmallRNAMappedPositionBuilder.Build(features.GroupByName().GroupBySequence(), Path.GetFileNameWithoutExtension(options.OutputFile), positionFile, m => m[0].Name.StringAfter(":"), positionByPercentage, minimumReadCount);
//clear all mapping information
foreach (var read in notNTAreads)
{
foreach (var loc in read.Locations)
{
loc.Features.Clear();
}
}
foreach (var feature in features)
{
feature.SamLocations.Clear();
}
}
}
private void WriteGroups(List <string> result, string resultFilename, List <FeatureItemGroup> allmapped, List <FeatureItem> otherFeatures, string biotype)
{
List <FeatureItemGroup> groups;
FeatureItemGroupCountWriter writer;
string displayBiotype;
if (string.IsNullOrEmpty(biotype))
{
displayBiotype = "other";
groups = otherFeatures.GroupByIdenticalQuery();
writer = new FeatureItemGroupCountWriter(m => m.DisplayName);
}
else
{
displayBiotype = biotype;
groups = otherFeatures.Where(m => m.Name.StartsWith(biotype)).GroupByIdenticalQuery();
writer = new FeatureItemGroupCountWriter(m => m.DisplayNameWithoutCategory);
}
if (groups.Count > 0)
{
OrderFeatureItemGroup(groups);
Progress.SetMessage("writing {0} count ...", displayBiotype);
var file = Path.ChangeExtension(resultFilename, "." + displayBiotype + ".count");
writer.WriteToFile(file, groups);
result.Add(file);
allmapped.AddRange(groups);
//if (!string.IsNullOrEmpty(biotype))
//{
var positionFile = Path.ChangeExtension(options.OutputFile, displayBiotype + ".position");
SmallRNAMappedPositionBuilder.Build(groups, Path.GetFileNameWithoutExtension(options.OutputFile), positionFile, m => m[0].Name.StringAfter(":"), true);
var absolutePositionFile = Path.ChangeExtension(options.OutputFile, displayBiotype + ".absolutePosition");
SmallRNAMappedPositionBuilder.Build(groups, Path.GetFileNameWithoutExtension(options.OutputFile), absolutePositionFile, m => m[0].Name.StringAfter(":"), false);
//}
}
groups.Clear();
}
public override IEnumerable <string> Process()
{
var result = new List <string>();
//read regions
var featureLocations = options.GetSequenceRegions();
Progress.SetMessage("There are {0} coordinate entries", featureLocations.Count);
if (featureLocations.Count == 0)
{
throw new Exception(string.Format("No coordinate found in file {0}", options.CoordinateFile));
}
var fGroups = featureLocations.GroupBy(l => l.Category).OrderByDescending(l => l.Count()).ToList();
foreach (var fg in fGroups)
{
Console.WriteLine("{0} = {1}", fg.Key, fg.Count());
}
var featureChroms = new HashSet <string>(from feature in featureLocations
select feature.Seqname);
var resultFilename = options.OutputFile;
result.Add(resultFilename);
HashSet <string> cca = new HashSet <string>();
if (File.Exists(options.CCAFile))
{
cca = new HashSet <string>(File.ReadAllLines(options.CCAFile));
}
//parsing reads
List <QueryInfo> totalQueries;
var reads = ParseCandidates(options.InputFiles, resultFilename, out totalQueries);
if (reads.Count == 0)
{
throw new ArgumentException("No read found in file " + options.InputFiles.Merge(","));
}
HashSet <string> excludeQueries = new HashSet <string>();
if (!string.IsNullOrEmpty(options.ExcludeXml))
{
Progress.SetMessage("Excluding queries in {0} ...", options.ExcludeXml);
excludeQueries = new HashSet <string>(from q in MappedItemGroupXmlFileFormat.ReadQueries(options.ExcludeXml)
select q.StringBefore(SmallRNAConsts.NTA_TAG));
reads.RemoveAll(m => excludeQueries.Contains(m.Locations.First().Parent.Qname.StringBefore(SmallRNAConsts.NTA_TAG)));
Progress.SetMessage("Total candidate {0} for mapping ...", reads.Count);
}
var hasMicroRnaNTA = reads.Any(l => l.NTA.Length > 0);
var hasTrnaNTA = hasMicroRnaNTA || File.Exists(options.CCAFile);
if (!options.NoCategory)
{
//First of all, draw candidate mapping position graph
var miRNAPositionFile = Path.ChangeExtension(options.OutputFile, SmallRNAConsts.miRNA + ".candidates.position");
if (!options.NotOverwrite || !File.Exists(miRNAPositionFile))
{
Progress.SetMessage("Drawing microRNA candidates position pictures...");
var notNTAreads = hasMicroRnaNTA ? reads.Where(m => m.NTA.Length == 0).ToList() : reads;
DrawPositionImage(notNTAreads, featureLocations.Where(m => m.Category.Equals(SmallRNAConsts.miRNA)).ToList(), SmallRNABiotype.miRNA.ToString(), miRNAPositionFile);
}
}
var featureGroups = new List <FeatureItemGroup>();
var mappedfile = resultFilename + ".mapped.xml";
if (File.Exists(mappedfile) && options.NotOverwrite)
{
Progress.SetMessage("Reading mapped feature items...");
featureGroups = new FeatureItemGroupXmlFormat().ReadFromFile(mappedfile);
}
else
{
Progress.SetMessage("Mapping feature items...");
//mapping reads to features based on miRNA, tRNA, mt_tRNA and other smallRNA priority
MapReadToSequenceRegion(featureLocations, reads, cca, hasMicroRnaNTA, hasTrnaNTA);
var featureMapped = featureLocations.GroupByName();
featureMapped.RemoveAll(m => m.GetEstimatedCount() == 0);
featureMapped.ForEach(m => m.CombineLocations());
if (options.NoCategory)
{
featureGroups = featureMapped.GroupByIdenticalQuery();
}
else
{
var mirnaGroups = featureMapped.Where(m => m.Name.StartsWith(SmallRNAConsts.miRNA)).GroupBySequence();
if (mirnaGroups.Count > 0)
{
OrderFeatureItemGroup(mirnaGroups);
Progress.SetMessage("writing miRNA count ...");
var mirnaCountFile = Path.ChangeExtension(resultFilename, "." + SmallRNAConsts.miRNA + ".count");
new SmallRNACountMicroRNAWriter(options.Offsets).WriteToFile(mirnaCountFile, mirnaGroups);
result.Add(mirnaCountFile);
featureGroups.AddRange(mirnaGroups);
var positionFile = Path.ChangeExtension(options.OutputFile, SmallRNAConsts.miRNA + ".position");
SmallRNAMappedPositionBuilder.Build(mirnaGroups, Path.GetFileNameWithoutExtension(options.OutputFile), positionFile, m => m[0].Name.StringAfter(":"));
}
mirnaGroups.Clear();
var trnaCodeGroups = featureMapped.Where(m => m.Name.StartsWith(SmallRNAConsts.tRNA)).GroupByFunction(SmallRNAUtils.GetTrnaAnticodon, false);
if (trnaCodeGroups.Count > 0)
{
OrderFeatureItemGroup(trnaCodeGroups);
Progress.SetMessage("writing tRNA code count ...");
var trnaCodeCountFile = Path.ChangeExtension(resultFilename, "." + SmallRNAConsts.tRNA + ".count");
new FeatureItemGroupCountWriter(m => m.DisplayNameWithoutCategory).WriteToFile(trnaCodeCountFile, trnaCodeGroups);
result.Add(trnaCodeCountFile);
featureGroups.AddRange(trnaCodeGroups);
var positionFile = Path.ChangeExtension(options.OutputFile, SmallRNAConsts.tRNA + ".position");
SmallRNAMappedPositionBuilder.Build(trnaCodeGroups, Path.GetFileName(options.OutputFile), positionFile, m => m[0].Name.StringAfter(":"));
}
trnaCodeGroups.Clear();
var otherFeatures = featureMapped.Where(m => !m.Name.StartsWith(SmallRNAConsts.miRNA) && !m.Name.StartsWith(SmallRNAConsts.tRNA)).ToList();
var exportBiotypes = SmallRNAUtils.GetOutputBiotypes(options);
foreach (var biotype in exportBiotypes)
{
WriteGroups(result, resultFilename, featureGroups, otherFeatures, biotype);
}
var leftFeatures = otherFeatures.Where(l => !exportBiotypes.Any(b => l.Name.StartsWith(b))).ToList();
WriteGroups(result, resultFilename, featureGroups, leftFeatures, null);
}
Progress.SetMessage("writing all smallRNA count ...");
new FeatureItemGroupCountWriter().WriteToFile(resultFilename, featureGroups);
result.Add(resultFilename);
Progress.SetMessage("writing mapping details...");
new FeatureItemGroupXmlFormatHand().WriteToFile(mappedfile, featureGroups);
}
var readSummary = GetReadSummary(featureGroups, excludeQueries, reads, totalQueries);
WriteInfoFile(result, resultFilename, readSummary, featureGroups);
result.Add(mappedfile);
Progress.End();
return(result);
}