public static bool IsPaired(string bamFile)
{
using (var sr = new FastqItemBAMParser(bamFile))
{
FastqItem item;
var count = 0;
while ((item = sr.ParseNext()) != null)
{
if (item.Flags.HasFlag(SAMFlags.PairedRead))
{
return(true);
}
if (item.Qname.EndsWith("/1") || item.Qname.EndsWith("/2"))
{
return(true);
}
count++;
if (count > 100000)
{
break;
}
}
}
return(false);
}
public override IEnumerable <string> Process()
{
Progress.SetMessage("This single end bam file has been sorted by name, it will cost less time/memory to generate fastq files ...");
var output = _options.OutputPrefix + ".fastq";
if (!_options.UnGzipped)
{
output = output + ".gz";
}
var tmp = output + ".tmp";
using (var sw = StreamUtils.GetWriter(tmp, !_options.UnGzipped))
{
using (var sr = new FastqItemBAMParser(_options.InputFile))
{
string lastname = null;
FastqItem ss;
var count = 0;
var outputCount = 0;
while ((ss = sr.ParseNext()) != null)
{
count++;
if (count % 100000 == 0)
{
Progress.SetMessage("{0} reads", count);
if (Progress.IsCancellationPending())
{
throw new UserTerminatedException();
}
}
if (!ss.Qname.Equals(lastname))
{
ss.WriteFastq(sw);
lastname = ss.Qname;
outputCount++;
if (outputCount % 1000000 == 0)
{
GC.Collect();
GC.WaitForPendingFinalizers();
Progress.SetMessage(string.Format("{0} single reads processed, cost memory: {1} MB", outputCount, (GC.GetTotalMemory(true) / 1048576)));
}
}
}
}
}
if (File.Exists(output))
{
File.Delete(output);
}
File.Move(tmp, output);
return(new[] { output });
}
public static bool IsSortedByName(string bamFile)
{
using (var sr = new FastqItemBAMParser(bamFile))
{
var hd = sr.Header.RecordFields.FirstOrDefault(m => m.Typecode.Equals("HD"));
if (hd != null)
{
var so = hd.Tags.FirstOrDefault(m => m.Tag.Equals("SO"));
if (so != null)
{
return(so.Value.Equals("queryname"));
}
}
}
return(false);
}
public override IEnumerable <string> Process()
{
Progress.SetMessage("This single end bam file is not sorted by name, it will cost more time/memory to generate fastq files ...");
var output = _options.OutputPrefix + ".fastq";
if (!_options.UnGzipped)
{
output = output + ".gz";
}
var tmp = output + ".tmp";
using (var sw = StreamUtils.GetWriter(tmp, !_options.UnGzipped))
{
using (var sr = new FastqItemBAMParser(_options.InputFile))
{
FastqItem ss;
var count = 0;
while ((ss = sr.ParseNext()) != null)
{
count++;
if (count % 100000 == 0)
{
Progress.SetMessage("{0} reads", count);
if (Progress.IsCancellationPending())
{
throw new UserTerminatedException();
}
}
ss.WriteFastq(sw);
sr.IgnoreQuery.Add(ss.Qname);
}
}
}
if (File.Exists(output))
{
File.Delete(output);
}
File.Move(tmp, output);
return(new[] { output });
}
public override IProcessor GetProcessor(Bam2FastqProcessorOptions options)
{
if (FastqItemBAMParser.IsPaired(options.InputFile))
{
if (FastqItemBAMParser.IsSortedByName(options.InputFile))
{
return(new Bam2PairedFastqNameSortedProcessor(options));
}
return(new Bam2PairedFastqProcessor(options));
}
if (FastqItemBAMParser.IsSortedByName(options.InputFile))
{
return(new Bam2SingleFastqNameSortedProcessor(options));
}
return(new Bam2SingleFastqProcessor(options));
}
public override IEnumerable <string> Process()
{
Progress.SetMessage("This pair end bam file has been sorted by name, it will cost less time to generate fastq files ...");
var output1 = _options.OutputPrefix + ".1.fastq";
var output2 = _options.OutputPrefix + ".2.fastq";
if (!_options.UnGzipped)
{
output1 = output1 + ".gz";
output2 = output2 + ".gz";
}
var tmp1 = output1 + ".tmp";
var tmp2 = output2 + ".tmp";
var output3 = Path.ChangeExtension(_options.OutputPrefix, ".orphan.fastq");
var paired = new FastqItem[3];
string lastname = null;
using (var sw1 = StreamUtils.GetWriter(tmp1, !_options.UnGzipped))
{
using (var sw2 = StreamUtils.GetWriter(tmp2, !_options.UnGzipped))
{
using (var sw3 = new StreamWriter(output3))
{
var sw = new[] { null, sw1, sw2, sw3 };
using (var sr = new FastqItemBAMParser(_options.InputFile))
{
FastqItem ss;
var count = 0;
var outputCount = 0;
while ((ss = sr.ParseNext()) != null)
{
if (string.IsNullOrEmpty(ss.Qname))
{
throw new Exception(string.Format("Entry after {0} has empty name", lastname));
}
ss.CheckPairedName();
if (string.IsNullOrEmpty(ss.Qname))
{
throw new Exception(string.Format("After check paired name, entry after {0} has empty name", lastname));
}
count++;
if (count % 100000 == 0)
{
Progress.SetMessage("{0} reads processed.", count);
if (Progress.IsCancellationPending())
{
throw new UserTerminatedException();
}
}
if (lastname == null)
{
paired[ss.PairIndex] = ss;
lastname = ss.PairName;
continue;
}
if (ss.PairName.Equals(lastname))
{
paired[ss.PairIndex] = ss;
continue;
}
outputCount = WriteFastq(paired, sw, outputCount);
paired[1] = null;
paired[2] = null;
paired[ss.PairIndex] = ss;
lastname = ss.PairName;
}
WriteFastq(paired, sw, outputCount);
}
}
}
}
if (File.Exists(output1))
{
File.Delete(output1);
}
File.Move(tmp1, output1);
if (File.Exists(output2))
{
File.Delete(output2);
}
File.Move(tmp2, output2);
if (new FileInfo(output3).Length == 0)
{
File.Delete(output3);
}
return(new[] { output1, output2 });
}
public override IEnumerable <string> Process()
{
Progress.SetMessage("This pair end bam file is not sorted by name, it will cost more time/memory to generate fastq files ...");
var map = new Dictionary <string, FastqItem>();
var output1 = _options.OutputPrefix + ".1.fastq";
var output2 = _options.OutputPrefix + ".2.fastq";
if (!_options.UnGzipped)
{
output1 = output1 + ".gz";
output2 = output2 + ".gz";
}
var tmp1 = output1 + ".tmp";
var tmp2 = output2 + ".tmp";
using (var sw1 = StreamUtils.GetWriter(tmp1, !_options.UnGzipped))
{
using (var sw2 = StreamUtils.GetWriter(tmp2, !_options.UnGzipped))
{
var sw = new[] { null, sw1, sw2 };
using (var sr = new FastqItemBAMParser(_options.InputFile))
{
FastqItem ss;
var count = 0;
var outputCount = 0;
while ((ss = sr.ParseNext()) != null)
{
ss.CheckPairedName();
count++;
if (count % 100000 == 0)
{
Progress.SetMessage("{0} reads processed, {1} unpaired.", count, map.Count);
if (Progress.IsCancellationPending())
{
throw new UserTerminatedException();
}
}
FastqItem paired;
if (map.TryGetValue(ss.PairName, out paired))
{
if (paired.PairIndex == ss.PairIndex)
{
continue;
}
ss.WriteFastq(sw[ss.PairIndex]);
paired.WriteFastq(sw[paired.PairIndex]);
sr.IgnoreQuery.Add(ss.Qname);
sr.IgnoreQuery.Add(paired.Qname);
map.Remove(ss.PairName);
outputCount++;
if (outputCount % 100000 == 0)
{
var temp = new Dictionary <string, FastqItem>(map);
map.Clear();
map = temp;
GC.Collect();
GC.WaitForPendingFinalizers();
Progress.SetMessage("Cost memory: " + (GC.GetTotalMemory(true) / 1048576) + " MB");
}
}
else
{
map[ss.PairName] = ss;
}
}
}
}
}
if (File.Exists(output1))
{
File.Delete(output1);
}
File.Move(tmp1, output1);
if (File.Exists(output2))
{
File.Delete(output2);
}
File.Move(tmp2, output2);
if (map.Count > 0)
{
var output3 = Path.ChangeExtension(_options.OutputPrefix, ".orphan.fastq");
using (var sw3 = new StreamWriter(output3))
{
foreach (var v in map.Values)
{
v.WriteFastq(sw3);
}
}
}
return(new[] { output1, output2 });
}
