public void TestReadFromFileWithoutFilter()
{
var data = new MpileupFisherResultFileFormat().ReadFromFile(@"../../../data/TCGA-BH-A0E0-DNA-TP-NB.candidates");
Assert.AreEqual(850, data.Count);
var res = data[0];
Assert.AreEqual("1", res.Item.SequenceIdentifier);
Assert.AreEqual(5732744, res.Item.Position);
Assert.AreEqual('G', res.Item.Nucleotide);
Assert.AreEqual("G", res.Group.SucceedName);
Assert.AreEqual("T", res.Group.FailedName);
Assert.AreEqual(44, res.Group.Sample1.Succeed);
Assert.AreEqual(1, res.Group.Sample1.Failed);
Assert.AreEqual(11, res.Group.Sample2.Succeed);
Assert.AreEqual(3, res.Group.Sample2.Failed);
Assert.AreEqual(3.8E-02, res.Group.PValue, 0.01);
}
public void TestReadFromFileWithFilter()
{
var data = new MpileupFisherResultFileFormat().ReadFromFile(@"../../../data/glmvc.candidates");
Assert.AreEqual(24, data.Count);
var res = data[3];
Assert.AreEqual("2", res.Item.SequenceIdentifier);
Assert.AreEqual(110301779, res.Item.Position);
Assert.AreEqual('A', res.Item.Nucleotide);
Assert.AreEqual("A", res.Group.SucceedName);
Assert.AreEqual("", res.Group.FailedName);
Assert.AreEqual(23, res.Group.Sample1.Succeed);
Assert.AreEqual(0, res.Group.Sample1.Failed);
Assert.AreEqual(9, res.Group.Sample2.Succeed);
Assert.AreEqual(0, res.Group.Sample2.Failed);
Assert.AreEqual(1.0E00, res.Group.PValue, 0.01);
Assert.AreEqual("Read depth < 10", res.FailedReason);
}
public override IEnumerable<string> Process()
{
if (!File.Exists(_options.BaseFilename) || new FileInfo(_options.BaseFilename).Length == 0)
{
base.Process();
}
else
{
Progress.SetMessage("Base file {0} exists, ignore pileup ...", _options.BaseFilename);
}
var filterOptions = options.GetFilterOptions();
if (new FileInfo(_options.BaseFilename).Length > 0)
{
if (!filterOptions.PrepareOptions())
{
throw new Exception("Filter options failed: " + filterOptions.ParsingErrors.Merge("\n"));
}
new FilterProcessor(filterOptions).Process();
var lines = File.ReadAllLines(filterOptions.ROutputFile).Skip(1).ToArray();
var glmfailed = lines.Count(m => m.Contains("GLM_PVALUE"));
var summarylines = File.ReadAllLines(_options.SummaryFilename).ToList();
if (summarylines.Last().StartsWith("glm pvalue"))
{
summarylines.RemoveAt(summarylines.Count - 1);
}
summarylines.Add(string.Format("glm pvalue > {0}\t{1}\t{2}", options.GlmPvalue, glmfailed, lines.Length - glmfailed));
File.WriteAllLines(_options.SummaryFilename, summarylines);
}
var mutationList = GetValidationList();
var candidates = new MpileupFisherResultFileFormat().ReadFromFile(options.CandidatesFilename).ToDictionary(m => GenomeUtils.GetKey(m.Item.SequenceIdentifier, m.Item.Position));
var items = new FilterItemTextFormat().ReadFromFile(filterOptions.ROutputFile).ToDictionary(m => GenomeUtils.GetKey(m.Chr, m.Start));
var result = new List<FilterItem>();
foreach (var mutation in mutationList.Items)
{
var key = GenomeUtils.GetKey(mutation.Chr, mutation.Pos);
if (items.ContainsKey(key))
{
result.Add(items[key]);
}
else
{
var item = new FilterItem();
item.Chr = mutation.Chr;
item.Start = mutation.Pos;
item.End = item.Start;
item.FisherNormal = string.Empty;
item.BrglmConverged = string.Empty;
item.BrglmGroup = 1.0;
item.BrglmGroupFdr = 1.0;
item.BrglmScore = string.Empty;
item.BrglmStrand = string.Empty;
item.BrglmPosition = string.Empty;
item.Identity = string.Empty;
result.Add(item);
if (candidates.ContainsKey(key))
{
var cand = candidates[key];
item.ReferenceAllele = cand.Item.Nucleotide.ToString();
item.MajorAllele = cand.Group.SucceedName;
item.MinorAllele = cand.Group.FailedName;
item.NormalMajorCount = cand.Group.Sample1.Succeed;
item.NormalMinorCount = cand.Group.Sample1.Failed;
item.TumorMajorCount = cand.Group.Sample2.Succeed;
item.TumorMinorCount = cand.Group.Sample2.Failed;
item.FisherGroup = cand.Group.PValue;
item.Filter = cand.FailedReason;
Console.WriteLine("In candidates : " + item.Filter);
}
else
{
item.NormalMajorCount = 0;
item.NormalMinorCount = 0;
item.TumorMajorCount = 0;
item.TumorMinorCount = 0;
item.Filter = "No coverage";
Console.WriteLine("No read : " + item.Filter);
}
}
}
new FilterItemVcfWriter(filterOptions).WriteToFile(_options.OutputSuffix + ".vcf", result);
new FilterItemTextFormat().WriteToFile(_options.OutputSuffix + ".tsv", result);
return new string[] { _options.OutputSuffix + ".tsv", _options.OutputSuffix + ".vcf" };
}