public void ComputePercentiles_Nominal()
{
var repeatNumbers = new[] { 10, 15, 20, 100, 200 };
var alleleCounts = new[] { 550, 34, 78, 30, 45 };
double[] expectedPercentiles =
{
0, 550 * 100.0 / 737, (550 + 34) * 100.0 / 737, (550 + 34 + 78) * 100.0 / 737,
(550 + 34 + 78 + 30) * 100.0 / 737
};
double[] observedResults = PercentileUtilities.ComputePercentiles(repeatNumbers.Length, alleleCounts);
Assert.Equal(expectedPercentiles, observedResults);
}
private static (ushort RefIndex, Interval <RepeatExpansionPhenotype> Interval) GetPhenotype(string line, IDictionary <string, IChromosome> refNameToChromosome)
{
string[] cols = line.OptimizedSplit('\t');
if (cols.Length < MinNumberOfColumns)
{
throw new InvalidDataException($"Expected at least {MinNumberOfColumns} columns in the STR data file, but found only {cols.Length}.");
}
string chromosomeString = cols[ChromIndex];
int start = int.Parse(cols[StartIndex]);
int end = int.Parse(cols[EndIndex]);
string phenotype = cols[PhenotypeIndex];
string omimId = cols[OmimIndex];
int[] repeatNumbers = cols[RepeatNumbersIndex].Split(',').Select(int.Parse).ToArray();
int[] alleleCounts = cols[AlleleCountsIndex].Split(',').Select(int.Parse).ToArray();
string[] classifications = cols[CategoriesIndex].Split(',').ToArray();
Interval[] classificationRanges = cols[CategoryRangesIndex].Split(',').Select(GetInterval).ToArray();
if (repeatNumbers.Length != alleleCounts.Length)
{
throw new InvalidDataException($"Inconsistent number of repeat numbers ({repeatNumbers.Length}) vs. allele counts ({alleleCounts.Length})");
}
if (classifications.Length != classificationRanges.Length)
{
throw new InvalidDataException($"Inconsistent number of values of classifications ({classifications.Length}) vs. classification ranges ({classificationRanges.Length})");
}
var chromosome = ReferenceNameUtilities.GetChromosome(refNameToChromosome, chromosomeString);
var chromosomeInterval = new ChromosomeInterval(chromosome, start, end);
double[] percentiles = PercentileUtilities.ComputePercentiles(repeatNumbers.Length, alleleCounts);
var rePhenotype = new RepeatExpansionPhenotype(chromosomeInterval, phenotype, omimId, repeatNumbers, percentiles, classifications, classificationRanges);
return(chromosome.Index, new Interval <RepeatExpansionPhenotype>(start, end, rePhenotype));
}
public void GetPercentile_RepeatNumberOutOfRange_NegativeIndex()
{
double observedResult = PercentileUtilities.GetPercentile(20, _values, _percentiles);
Assert.Equal(100, observedResult);
}
public void GetPercentile_RepeatNumberInRange_PositiveIndex()
{
double observedResult = PercentileUtilities.GetPercentile(14, _values, _percentiles);
Assert.Equal(99.5, observedResult);
}
