private static AlleleSet GenerateAlleleSet(PositionSet positionSet)
{
var alleleArrays = new string[positionSet._numPositions][];
var starts = positionSet.SimplePositions.Select(x => x.Start).ToArray();
for (var index = 0; index < positionSet._numPositions; index++)
{
var position = positionSet.SimplePositions[index];
alleleArrays[index] = new string[position.AltAlleles.Length + 1];
alleleArrays[index][0] = position.RefAllele;
position.AltAlleles.CopyTo(alleleArrays[index], 1);
}
return(new AlleleSet(positionSet.SimplePositions[0].Chromosome, starts, alleleArrays));
}
private static SampleInfo GetSampleInfo(PositionSet positionSet)
{
var sampleInfo = new string[positionSet._numPositions, positionSet.NumSamples][];
for (var i = 0; i < positionSet._numPositions; i++)
{
for (int sampleIndex = 0; sampleIndex < positionSet.NumSamples; sampleIndex++)
{
int sampleColIndex = sampleIndex + VcfCommon.GenotypeIndex;
sampleInfo[i, sampleIndex] = positionSet.SimplePositions[i].VcfFields[sampleColIndex].OptimizedSplit(':');
}
}
return(new SampleInfo(sampleInfo));
}
// GenotypeBlocks can be shared by multiple samples
// We mainly utilize phase set information at this step to avoid duplicated calculation
// These GenotypeBlocks could be further segmented when more details considered
private static IEnumerable <GenotypeBlock> GetGenotypeBlocks(PositionSet positionSet, int sampleIndex)
{
var genotypes = positionSet.GtInfo.Values[sampleIndex];
var entireBlock = new GenotypeBlock(genotypes);
var blockRanges = GetGenotypeBlockRange(positionSet.PsInfo.Values[sampleIndex], genotypes.Select(x => x.IsPhased).ToArray(), genotypes.Select(x => x.IsHomozygous).ToArray());
var genotypeBlocks = new List <GenotypeBlock>();
foreach (var range in blockRanges)
{
genotypeBlocks.Add(entireBlock.GetSubBlock(range.StartIndex, range.PositionCount));
}
return(genotypeBlocks);
}
private static HashSet <int>[] GetAllelesWithUnsupportedTypes(PositionSet positionSet)
{
var allelesWithUnsupportedTypes = new HashSet <int> [positionSet._numPositions];
for (int posIndex = 0; posIndex < positionSet._numPositions; posIndex++)
{
allelesWithUnsupportedTypes[posIndex] = new HashSet <int>();
var thisPosition = positionSet.SimplePositions[posIndex];
for (int varIndex = 0; varIndex < thisPosition.AltAlleles.Length; varIndex++)
{
if (!(IsSupportedVariantType(thisPosition.RefAllele, thisPosition.AltAlleles[varIndex]) || thisPosition.VcfFields[VcfCommon.AltIndex] == VcfCommon.GatkNonRefAllele))
{
allelesWithUnsupportedTypes[posIndex].Add(varIndex + 1); // GT tag is 1-based
}
}
}
return(allelesWithUnsupportedTypes);
}
public static PositionSet CreatePositionSet(List <ISimplePosition> simpleSimplePositions, List <int> functionBlockRanges)
{
var positionSet = new PositionSet(simpleSimplePositions, functionBlockRanges);
positionSet.AlleleSet = GenerateAlleleSet(positionSet);
positionSet._allelesWithUnsupportedTypes = GetAllelesWithUnsupportedTypes(positionSet);
positionSet._sampleInfo = GetSampleInfo(positionSet);
var phaseSetAndGqIndexes = positionSet.GetSampleTagIndexes(new[] { "GT", "PS", "GQ" });
positionSet.GtInfo = TagInfo <Genotype> .GetTagInfo(positionSet._sampleInfo, phaseSetAndGqIndexes[0], ExtractSampleValue, Genotype.GetGenotype);
positionSet.PsInfo = TagInfo <string> .GetTagInfo(positionSet._sampleInfo, phaseSetAndGqIndexes[1], ExtractSampleValue, x => x);
positionSet.GqInfo = TagInfo <string> .GetTagInfo(positionSet._sampleInfo, phaseSetAndGqIndexes[2], ExtractSampleValue, x => x);
var genotypeToSampleIndex = GetGenotypeToSampleIndex(positionSet);
var alleleBlockToSampleHaplotype = AlleleBlock.GetAlleleBlockToSampleHaplotype(genotypeToSampleIndex, positionSet._allelesWithUnsupportedTypes, positionSet.AlleleSet.Starts, positionSet.FunctionBlockRanges, out var alleleBlockGraph);
positionSet.AlleleBlockToSampleHaplotype = AlleleBlockMerger.Merge(alleleBlockToSampleHaplotype, alleleBlockGraph);
return(positionSet);
}
private static Dictionary <GenotypeBlock, List <int> > GetGenotypeToSampleIndex(PositionSet positionSet)
{
var genotypeToSample = new Dictionary <GenotypeBlock, List <int> >();
for (int sampleIndex = 0; sampleIndex < positionSet.NumSamples; sampleIndex++)
{
var genotypesAndStartIndexes = GetGenotypeBlocks(positionSet, sampleIndex);
foreach (var genotypeAndStartIndex in genotypesAndStartIndexes)
{
if (genotypeToSample.ContainsKey(genotypeAndStartIndex))
{
genotypeToSample[genotypeAndStartIndex].Add(sampleIndex);
}
else
{
genotypeToSample[genotypeAndStartIndex] = new List <int> {
sampleIndex
}
};
}
}
return(genotypeToSample);
}