public static (ImmutableDictionary <IChromosome, List <int> > PositionsByChromosome, int Count) GetPositions(Stream vcfStream, GenomicRange genomicRange,
ISequenceProvider sequenceProvider, IRefMinorProvider refMinorProvider)
{
var positionsByChromosome = new Dictionary <IChromosome, List <int> >();
var rangeChecker = new GenomicRangeChecker(genomicRange);
var refNameToChrom = sequenceProvider.RefNameToChromosome;
using (var reader = new StreamReader(vcfStream))
{
string line;
string currentReferenceName = "";
IChromosome chromosome = null;
while ((line = reader.ReadLine()) != null)
{
if (line.StartsWith('#'))
{
continue;
}
string[] cols = line.OptimizedSplit('\t');
string referenceName = cols[VcfCommon.ChromIndex];
if (referenceName != currentReferenceName)
{
if (!refNameToChrom.TryGetValue(referenceName, out chromosome))
{
continue;
}
currentReferenceName = referenceName;
}
(int position, bool foundError) = cols[VcfCommon.PosIndex].OptimizedParseInt32();
if (foundError)
{
throw new InvalidDataException($"Unable to convert the VCF position to an integer: {cols[VcfCommon.PosIndex]}");
}
if (rangeChecker.OutOfRange(chromosome, position))
{
break;
}
string refAllele = cols[VcfCommon.RefIndex];
string altAllele = cols[VcfCommon.AltIndex];
if (altAllele == "." && !IsRefMinor(refMinorProvider, chromosome, position))
{
continue;
}
sequenceProvider.LoadChromosome(chromosome);
TryAddPosition(positionsByChromosome, chromosome, position, refAllele, altAllele, sequenceProvider.Sequence);
}
}
int count = SortPositionsAndGetCount(positionsByChromosome);
return(positionsByChromosome.ToImmutableDictionary(), count);
}
public static IDictionary <IChromosome, List <int> > GetPositions(Stream vcfStream, GenomicRange genomicRange, ISequenceProvider sequenceProvider)
{
var benchmark = new Benchmark();
Console.Write("Scanning positions required for SA pre-loading....");
var chromPositions = new Dictionary <IChromosome, List <int> >();
var rangeChecker = new GenomicRangeChecker(genomicRange);
var refNameToChrom = sequenceProvider.RefNameToChromosome;
using (var reader = new StreamReader(vcfStream))
{
string line;
while ((line = reader.ReadLine()) != null)
{
if (!NeedProcessThisLine(refNameToChrom, line, out var splits, out IChromosome iChrom))
{
continue;
}
int position = int.Parse(splits[VcfCommon.PosIndex]);
if (rangeChecker.OutOfRange(iChrom, position))
{
break;
}
string refAllele = splits[VcfCommon.RefIndex];
string altAllele = splits[VcfCommon.AltIndex];
sequenceProvider.LoadChromosome(iChrom);
UpdateChromToPositions(chromPositions, iChrom, position, refAllele, altAllele, sequenceProvider.Sequence);
}
}
int count = SortPositionsAndGetCount(chromPositions);
Console.WriteLine($"{count} positions found in {Benchmark.ToHumanReadable(benchmark.GetElapsedTime())}");
return(chromPositions);
}
public bool PassedTheEnd(IChromosome chromosome, int position) => _genomicRangeChecker.OutOfRange(chromosome, position);