private VariantCall[] LoadIlluminaMatrixReportVariant(string line)
{
if (line[0] != 'r')
{
return null;
}
if (line[1] != 's')
{
return null;
}
if (_splitter == null)
{
var testSplit = line.Split('\t');
_splitter = new StringSplitter(testSplit.Length);
}
_splitter.Split(line, '\t');
var lineSplit = _splitter.Results;
var variants = new VariantCall[lineSplit.Length - 1];
for (var i = 1; i < lineSplit.Length; i++)
{
var variant = new VariantCall();
variant.Id = lineSplit[0];
variant.AlleleA = lineSplit[i][0].ToString();
variant.AlleleB = lineSplit[i][1].ToString();
variants[i - 1] = variant;
}
return variants;
}
public void LoadUcscSnps(string snpPath, bool clearSnps, IProgressReporter progress = null)
{
//Clear?
if (clearSnps)
{
Snps.Clear();
}
//Log
progress?.Update("Loading SNPs");
//Init
var lineCount = 0;
_splitter = null;
//Read file
using (var reader = _fileReaderFactory.GetFileReader(snpPath))
{
foreach (var line in reader)
{
//Loop init
lineCount++;
if (lineCount == 1)
{
var testSplit = line.Split('\t');
_splitter = new StringSplitter(testSplit.Length);
continue;
}
//Read data
// ReSharper disable once PossibleNullReferenceException
_splitter.Split(line, '\t');
var lineSplit = _splitter.Results;
var snp = new Snp();
//Build data
snp.Chromosome = lineSplit[1];
snp.Pos = lineSplit[2];
snp.RsId = lineSplit[4];
snp.Strand = lineSplit[6];
snp.SnpRef = lineSplit[7];
snp.Observed = lineSplit[9];
if (snp.SnpRef.Length == 1 && snp.SnpRef != "-")
{
if (snp.Strand == "+")
{
snp.SnpAlt = snp.Observed[0] == snp.SnpRef[0] ? snp.Observed[2].ToString() : snp.Observed[0].ToString();
}
else
{
var newObs = new char[3];
newObs[1] = '/';
//Reverse observed
if(snp.Observed[0] == 'A') { newObs[2] = 'T';}
else if (snp.Observed[0] == 'T') { newObs[2] = 'A'; }
else if (snp.Observed[0] == 'G') { newObs[2] = 'C'; }
else if (snp.Observed[0] == 'C') { newObs[2] = 'G'; }
if (snp.Observed[2] == 'A') { newObs[0] = 'T'; }
else if (snp.Observed[2] == 'T') { newObs[0] = 'A'; }
else if (snp.Observed[2] == 'G') { newObs[0] = 'C'; }
else if (snp.Observed[2] == 'C') { newObs[0] = 'G'; }
var observed = new String(newObs);
snp.SnpAlt = observed[0] == snp.SnpRef[0] ? observed[2].ToString() : observed[0].ToString();
}
if (!Snps.ContainsKey(snp.RsId))
{
Snps.Add(snp.RsId, snp);
}
}
//Progress
lineCount++;
if (progress != null && lineCount % 10 == 1)
{
var perc = (double)reader.Stream.Position / reader.Stream.Length;
progress.Update(perc);
}
}
}
}
private VariantCall[] LoadVcfVariant(string line)
{
//Check for comment headers
if (line.Length < 2)
{
return null;
}
if (line.Substring(0, 2) == "##")
{
return null;
}
if (line[0] == '#')
{
return null;
}
if (_splitter == null)
{
var testSplit = line.Split('\t');
_splitter = new StringSplitter(testSplit.Length);
}
_splitter.Split(line, '\t');
var lineSplit = _splitter.Results;
//Check if we passed QC
if (lineSplit[6] != "PASS")
{
return null;
}
//Check if id starts with rs
if (lineSplit[2][0] != 'r')
{
return null;
}
if (lineSplit[2][1] != 's')
{
return null;
}
//Create variant
var newVar = new VariantCall();
newVar.Id = lineSplit[2];
newVar.VcfRef = lineSplit[3];
newVar.VcfAlt = lineSplit[4];
return new[] {newVar};
}
private VariantCall[] LoadIlluminaFinalReportVariant(string line)
{
if (line[0] != 'r')
{
return null;
}
if (line[1] != 's')
{
return null;
}
if (_splitter == null)
{
var testSplit = line.Split('\t');
_splitter = new StringSplitter(testSplit.Length);
}
_splitter.Split(line, '\t');
var lineSplit = _splitter.Results;
var newVar = new VariantCall();
newVar.Id = lineSplit[0];
newVar.AlleleA = lineSplit[16];
newVar.AlleleB = lineSplit[17];
return new[] { newVar };
}
public void AnalyseVcfFile(string variantPath, VariantFileType fileType, string snpPath = null,
ConsoleLogger logger = null)
{
ProgressReporter pr = null;
logger?.LogInfo(LogName, "Analysing variants");
//Do we need to load snps?
if (snpPath != null)
{
pr = null;
if (logger != null)
{
pr = new ProgressReporter(logger, 1);
}
logger?.StartTask(1, "Load SNP file");
LoadUcscSnps(snpPath, true, pr);
logger?.EndTask(1);
logger?.LogInfo(LogName, "Loaded " + Snps.Count + " SNPs");
}
//Init
_splitter = null;
var variantCount = 0;
var unmatchedCount = 0;
//Var vcf
var calledForwardOnPlus = 0;
var calledForwardOnNeg = 0;
//Var final report
var matchedFwdStrand = 0;
var unMatchedFwdStrand = 0;
//Logging
pr = null;
if (logger != null)
{
pr = new ProgressReporter(logger, 1);
}
logger?.StartTask(1, "Processing variants");
//Read file
using (var reader = _fileReaderFactory.GetFileReader(variantPath))
{
foreach (var line in reader)
{
//Load variants
var variants = LoadVariant(fileType, line);
//Continue if not a real variants
if (variants == null)
{
continue;
}
for (var i = 0; i < variants.Length; i++)
{
//Get current variant
var variant = variants[i];
//Try and match to snp's loads
Snp matchSnp;
Snps.TryGetValue(variant.Id, out matchSnp);
//Check we have a match
if (matchSnp != null)
{
if (fileType == VariantFileType.Vcf)
{
if (matchSnp.Strand == "+" && matchSnp.SnpRef == variant.VcfRef)
{
calledForwardOnPlus++;
}
else if (matchSnp.Strand == "-" && matchSnp.SnpRef == variant.VcfRef)
{
calledForwardOnNeg++;
}
}
else if (fileType == VariantFileType.IlluminiaFinalReport)
{
if (variant.AlleleA != "-" && variant.AlleleB != "-")
{
if ((variant.AlleleA == matchSnp.SnpRef || variant.AlleleA == matchSnp.SnpAlt) &&
(variant.AlleleB == matchSnp.SnpRef || variant.AlleleB == matchSnp.SnpAlt))
{
matchedFwdStrand++;
}
else
{
if (matchSnp.Observed.Length <= 3)
{
unMatchedFwdStrand++;
}
}
}
}
else if (fileType == VariantFileType.IlluminaMatrix)
{
if (variant.AlleleA != "-" && variant.AlleleB != "-")
{
if ((variant.AlleleA == matchSnp.SnpRef || variant.AlleleA == matchSnp.SnpAlt) &&
(variant.AlleleB == matchSnp.SnpRef || variant.AlleleB == matchSnp.SnpAlt))
{
matchedFwdStrand++;
}
else
{
if (matchSnp.Observed.Length <= 3)
{
unMatchedFwdStrand++;
}
}
}
}
}
else
{
unmatchedCount++;
}
variantCount++;
if (logger != null)
{
var perc = (double) reader.Stream.Position/reader.Stream.Length;
pr.Update(perc);
if (fileType == VariantFileType.Vcf)
{
pr.Update(
$" TV:{variantCount} UTV:{unmatchedCount} :CallFwd+:{calledForwardOnPlus} CallFwd-:{calledForwardOnNeg}");
}
else if (fileType == VariantFileType.IlluminiaFinalReport)
{
pr.Update(
$" TV:{variantCount} UTV:{unmatchedCount} MatchedFwd:{matchedFwdStrand} UnMatchedFwd:{unMatchedFwdStrand}");
}
else if (fileType == VariantFileType.IlluminaMatrix)
{
pr.Update(
$" TV:{variantCount} UTV:{unmatchedCount} MatchedFwd:{matchedFwdStrand} UnMatchedFwd:{unMatchedFwdStrand}");
}
}
}
}
}
logger?.EndTask(1);
}