/// <summary>
/// returns a ClinVar object given the vcf line
/// </summary>
public static DgvItem ExtractDgvItem(string line, IDictionary <string, IChromosome> refChromDict)
{
var cols = line.Split('\t');
if (cols.Length < 8)
{
return(null);
}
var id = cols[0];
var chromosomeName = cols[1];
if (!refChromDict.ContainsKey(chromosomeName))
{
return(null);
}
var chromosome = refChromDict[chromosomeName];
var start = int.Parse(cols[2]);
var end = int.Parse(cols[3]);
var variantType = cols[4];
var variantSubType = cols[5];
var sampleSize = int.Parse(cols[14]);
var observedGains = cols[15] == "" ? 0 : int.Parse(cols[15]);
var observedLosses = cols[16] == "" ? 0 : int.Parse(cols[16]);
var seqAltType = SaParseUtilities.GetSequenceAlterationType(variantType, variantSubType);
return(new DgvItem(id, chromosome, start, end, sampleSize, observedGains, observedLosses, seqAltType));
}
private OnekGenSvItem ExtractOneKGenSvItem(string line)
{
var splitLine = line.OptimizedSplit('\t');
string altAllele = splitLine[AltIndex];
if (altAllele.StartsWith("<INS:ME:"))
{
return(null);
}
string chromosomeName = splitLine[ChromIndex];
if (!_refNameDict.ContainsKey(chromosomeName))
{
return(null);
}
var chromosome = _refNameDict[chromosomeName];
int start = int.Parse(splitLine[StartIndex]) + 1; // start is 0-based in BED format
int end = int.Parse(splitLine[EndIndex]);
string id = RemoveMissingValues(splitLine[IdIndex]);
string infoFields = splitLine[InfoIndex];
Clear();
ParseInfoField(infoFields);
var variantType = SaParseUtilities.GetSequenceAlteration(_svType);
return(new OnekGenSvItem(chromosome, start, end, variantType, id,
_allAlleleNumber, _allAlleleCount,
_allAlleleFrequency, _afrAlleleFrequency, _amrAlleleFrequency, _easAlleleFrequency, _eurAlleleFrequency, _sasAlleleFrequency));
}
private OnekGenSvItem ExtractOneKGenSvItem(string line)
{
var splitLine = line.OptimizedSplit('\t'); // we don't care about the many fields after info field
if (splitLine.Length < 8)
{
return(null);
}
var altAlleles = splitLine[VcfCommon.AltIndex].OptimizedSplit(',');
var hasSymbolicAllele = altAlleles.Any(x => x.OptimizedStartsWith('<') && x.OptimizedEndsWith('>'));
if (!hasSymbolicAllele)
{
return(null);
}
var chromosomeName = splitLine[VcfCommon.ChromIndex];
if (!_refNameDict.ContainsKey(chromosomeName))
{
return(null);
}
var chromosome = _refNameDict[chromosomeName];
var position = int.Parse(splitLine[VcfCommon.PosIndex]); //we have to get it from RSPOS in info
var id = splitLine[VcfCommon.IdIndex];
//var refAllele = splitLine[VcfCommon.RefIndex];
var infoFields = splitLine[VcfCommon.InfoIndex];
Clear();
ParseInfoField(infoFields);
if (_svEnd == null && _svLen != null)
{
_svEnd = position + _svLen;
}
if (_svEnd == null)
{
return(null);
}
var variantType = SaParseUtilities.GetSequenceAlteration(_svType);
return(new OnekGenSvItem(chromosome, position + 1, _svEnd.Value, variantType, id,
_allAlleleNumber, _allAlleleCount,
_allAlleleFrequency, _afrAlleleFrequency, _amrAlleleFrequency, _easAlleleFrequency, _eurAlleleFrequency, _sasAlleleFrequency));
}
public override SupplementaryIntervalItem GetSupplementaryInterval()
{
if (!IsInterval)
{
return(null);
}
var seqAltType = SaParseUtilities.GetSequenceAlteration(SvType, ObservedGains, ObservedLosses);
var intValues = new Dictionary <string, int>();
var doubleValues = new Dictionary <string, double>();
var freqValues = new Dictionary <string, double>();
var stringValues = new Dictionary <string, string>();
var boolValues = new List <string>();
var suppInterval = new SupplementaryIntervalItem(Chromosome, Start, SvEnd, AlternateAllele, seqAltType,
"1000 Genomes Project", intValues, doubleValues, freqValues, stringValues, boolValues);
if (Id != null)
{
suppInterval.AddStringValue("id", Id);
}
if (AfrFreq != null)
{
suppInterval.AddFrequencyValue("variantFreqAfr", Convert.ToDouble(AfrFreq));
}
if (AllFreq != null)
{
suppInterval.AddFrequencyValue("variantFreqAll", Convert.ToDouble(AllFreq));
}
if (AmrFreq != null)
{
suppInterval.AddFrequencyValue("variantFreqAmr", Convert.ToDouble(AmrFreq));
}
if (EasFreq != null)
{
suppInterval.AddFrequencyValue("variantFreqEas", Convert.ToDouble(EasFreq));
}
if (EurFreq != null)
{
suppInterval.AddFrequencyValue("variantFreqEur", Convert.ToDouble(EurFreq));
}
if (SasFreq != null)
{
suppInterval.AddFrequencyValue("variantFreqSas", Convert.ToDouble(SasFreq));
}
if (AllAlleleNumber != null && AllAlleleNumber.Value > 0)
{
suppInterval.AddIntValue("sampleSize", AllAlleleNumber.Value);
}
if (AfrAlleleNumber != null && AfrAlleleNumber.Value > 0)
{
suppInterval.AddIntValue("sampleSizeAfr", AfrAlleleNumber.Value);
}
if (AmrAlleleNumber != null && AmrAlleleNumber.Value > 0)
{
suppInterval.AddIntValue("sampleSizeAmr", AmrAlleleNumber.Value);
}
if (EasAlleleNumber != null && EasAlleleNumber.Value > 0)
{
suppInterval.AddIntValue("sampleSizeEas", EasAlleleNumber.Value);
}
if (EurAlleleNumber != null && EurAlleleNumber.Value > 0)
{
suppInterval.AddIntValue("sampleSizeEur", EurAlleleNumber.Value);
}
if (SasAlleleNumber != null && SasAlleleNumber.Value > 0)
{
suppInterval.AddIntValue("sampleSizeSas", SasAlleleNumber.Value);
}
if (ObservedGains != 0)
{
suppInterval.AddIntValue("observedGains", ObservedGains);
}
if (ObservedLosses != 0)
{
suppInterval.AddIntValue("observedLosses", ObservedLosses);
}
return(suppInterval);
}