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);
}
}
}
}