public VcfFilter(PsaraOptions settings)
{
var vcfIn = settings.VcfPath;
var vcfName = Path.GetFileName(vcfIn);
_originalHeaderLines = AlleleReader.GetAllHeaderLines(vcfIn);
_geometricFilter = new GeometricFilter(settings.GeometricFilterParameters);
_psaraOptions = (PsaraOptions)VcfConsumerAppParsingUtils.TryToUpdateWithOriginalOptions(settings, _originalHeaderLines, vcfIn);
_outputFile = Path.Combine(settings.OutputDirectory, vcfName.Replace(".vcf", ".filtered.vcf"));
_outputFile = _outputFile.Replace(".genome.filtered.vcf", ".filtered.genome.vcf");
}
public PsaraOptionsParser()
{
Options = new PsaraOptions();
}
public static void DoFiltering(PsaraOptions settings)
{
var geometricFilter = new GeometricFilter(settings.GeometricFilterParameters);
//maybe expand to add other filters..
var vcfIn = settings.InputVcf;
var vcfName = Path.GetFileName(vcfIn);
var outputFile = Path.Combine(settings.OutputDirectory, vcfName.Replace(".vcf", ".filtered.vcf"));
outputFile = outputFile.Replace(".genome.filtered.vcf", ".filtered.genome.vcf");
Logger.WriteToLog("filtering " + vcfIn + "...");
if (File.Exists(outputFile))
{
File.Delete(outputFile);
}
List <string> header = VcfReader.GetAllHeaderLines(vcfIn);
string cmdLine = "##Psara_cmdline=" + settings.QuotedCommandLineArgumentsString;
VcfWriterConfig config = GetWriterConfigToMatchInputVcf(vcfIn);
using (PsaraVcfWriter writer = new PsaraVcfWriter(outputFile, config, new VcfWriterInputContext(), header, cmdLine))
{
writer.WriteHeader();
using (VcfReader reader = new VcfReader(vcfIn, false))
{
var backLogVcfVariant = new VcfVariant();
var coLocatedAlleles = new List <CalledAllele>();
var moreVariantsInVcf = reader.GetNextVariant(backLogVcfVariant);
var incomingBatch = new List <CalledAllele>();
while (moreVariantsInVcf)
{
if (incomingBatch.Count == 0)
{
incomingBatch = moreVariantsInVcf ? VcfVariantUtilities.Convert(new List <VcfVariant> {
backLogVcfVariant
},
config.ShouldOutputRcCounts, config.ShouldOutputTsCounts, false).ToList() : null;
moreVariantsInVcf = reader.GetNextVariant(backLogVcfVariant);
}
if ((coLocatedAlleles.Count == 0) || AreColocated(coLocatedAlleles, incomingBatch))
{
coLocatedAlleles.AddRange(incomingBatch);
incomingBatch.Clear();
//colocated alleles are left behind
}
else
{
FilterAndStreamOut(coLocatedAlleles, writer, geometricFilter);
coLocatedAlleles.Clear();
//incomingBatch alleles are left behind
}
}
//if you get here, there is no more unprocessed vcf variants but there could be
//colocated or an incoming batch of alleles left over. We need to write them to file before exiting.
FilterAndStreamOut(coLocatedAlleles, writer, geometricFilter);
FilterAndStreamOut(incomingBatch, writer, geometricFilter);
}
}
}