public void ColumnsShuffle(byte chrColumn, byte leftColumn, sbyte rightColumn, byte nameColumn, byte valueColumn)
{
// Arrange
var rg = new RegionGenerator
{
ChrColumn = chrColumn,
LeftColumn = leftColumn,
RightColumn = rightColumn,
NameColumn = nameColumn,
ValueColumn = valueColumn
};
using (var file = new TempFileCreator(rg))
{
// Act
var parser = new BedParser(
new BedColumns()
{
Chr = chrColumn,
Left = leftColumn,
Right = rightColumn,
Name = nameColumn,
Value = valueColumn
});
var parsedPeak = parser.Parse(file.Path).Chromosomes[rg.Chr].Strands[rg.Strand].Intervals.ToList()[0];
// Assert
Assert.True(parsedPeak.CompareTo(rg.Peak) == 0);
}
}
public void TestMultiStrand()
{
// Arrange
string[] peaks = new string[]
{
"chr1\t10\t20\t*\tGeUtilities_00\t0.1",
"chr1\t30\t40\t+\tGeUtilities_01\t0.11",
"chr1\t50\t60\t-\tGeUtilities_02\t0.111",
"chr1\t50\t60\t#\tGeUtilities_02\t0.111", // Any strand name other than '+', '-', and '*' will be parsed as '*'.
};
using (var file = new TempFileCreator(peaks))
{
// Act
var parser = new BedParser(
new BedColumns
{
Chr = 0,
Left = 1,
Right = 2,
Strand = 3,
Name = 4,
Value = 5
});
var parsedData = parser.Parse(file.Path);
// Assert
Assert.True(parsedData.Chromosomes["chr1"].Strands.ContainsKey('*'));
Assert.True(parsedData.Chromosomes["chr1"].Strands.ContainsKey('+'));
Assert.True(parsedData.Chromosomes["chr1"].Strands.ContainsKey('-'));
Assert.False(parsedData.Chromosomes["chr1"].Strands.ContainsKey('#'));
}
}
public void TestMinMaxValueWhenDefaultValueIsUsed()
{
// Arrange
double defaultValue = 1e-8;
string[] peaks = new string[]
{
"chr1\t10\t20",
"chr1\t30\t40",
"chr1\t50\t60",
"chr1\t70\t80",
};
using (var file = new TempFileCreator(peaks))
{
// Act
var parser = new BedParser()
{
DropPeakIfInvalidValue = false,
DefaultValue = defaultValue
};
var parsedData = parser.Parse(file.Path);
// Assert
Assert.True(parsedData.PValueMax.Value == defaultValue);
Assert.True(parsedData.PValueMin.Value == defaultValue);
}
}
public void FailReadValue()
{
// Arrange
using (var file = new TempFileCreator("chr1\t10\t20\tGeUtilities_01\t123..45"))
{
// Act
var parser = new BedParser();
var parsedData = parser.Parse(file.Path);
// Assert
Assert.False(parsedData.Chromosomes.ContainsKey("chr1"));
}
}
public void AvoidEmptyLines()
{
// Arrange
using (var file = new TempFileCreator(" "))
{
// Act
var parser = new BedParser();
var parsedData = parser.Parse(file.Path);
// Assert
Assert.True(parsedData.Chromosomes.Count == 0);
}
}
public void DefaultColumnsOrder()
{
// Arrange
var rg = new RegionGenerator();
using (var file = new TempFileCreator())
{
// Act
var parser = new BedParser();
var parsedPeak = parser.Parse(file.Path).Chromosomes[rg.Chr].Strands[rg.Strand].Intervals.ToList()[0];
// Assert
Assert.True(parsedPeak.CompareTo(rg.Peak) == 0);
}
}
public void FullySetArguments()
{
// Arrange
var rg = new RegionGenerator();
using (var file = new TempFileCreator(rg))
{
// Act
var parser = new BedParser(rg.Columns);
var parsedPeak = parser.Parse(file.Path).Chromosomes[rg.Chr].Strands[rg.Strand].Intervals[0];
// Assert
Assert.True(parsedPeak.CompareTo(rg.Peak) == 0);
}
}
public void ReadStrand()
{
// Arrange
var rg = new RegionGenerator();
using (var file = new TempFileCreator(rg))
{
// ACt
var parser = new BedParser();
var parsedData = parser.Parse(file.Path);
// Assert
Assert.True(parsedData.Chromosomes[rg.Chr].Strands.ContainsKey(rg.Strand));
}
}
public void AssignHashKey()
{
// Arrange
var rg = new RegionGenerator();
using (var file = new TempFileCreator(rg))
{
// Act
var parser = new BedParser();
var parsedData = parser.Parse(file.Path);
// Assert
Assert.True(parsedData.Chromosomes[rg.Chr].Strands[rg.Strand].Intervals[0].GetHashCode() != 0);
}
}
public void FailReadChr(string chr)
{
// Arrange
using (var file = new TempFileCreator(new RegionGenerator {
Chr = "chr1"
}))
{
// Act
var parser = new BedParser();
var parsedData = parser.Parse(file.Path);
// Assert
Assert.False(parsedData.Chromosomes.ContainsKey(chr));
}
}
public void ReadValue()
{
// Arrange
var rg = new RegionGenerator();
using (var file = new TempFileCreator(rg))
{
// Act
var parser = new BedParser();
var parsedData = parser.Parse(file.Path);
// Assert
Assert.True(parsedData.Chromosomes[rg.Chr].Strands[rg.Strand].Intervals[0].Value == rg.Value);
}
}
public void FailReadRightColumnIndexOutOfRange()
{
// Arrange
using (var file = new TempFileCreator("chr1\t10\t20\tGeUtilities_01\t123.4"))
{
// Act
var parser = new BedParser(new BedColumns()
{
Right = 10
});
var parsedData = parser.Parse(file.Path);
// Assert
Assert.False(parsedData.Chromosomes.ContainsKey("chr1"));
}
}
public void AvoidHeader(int headerCount, byte readOffset)
{
// Arrange
using (var file = new TempFileCreator(new RegionGenerator(), headerLineCount: headerCount))
{
// Act
var parser = new BedParser()
{
ReadOffset = readOffset
};
var parsedData = parser.Parse(file.Path);
// Assert
Assert.True(parsedData.Chromosomes.Count == 1);
}
}
public void DropPeaksHavingInvalidPValue()
{
// Arrange
using (var file = new TempFileCreator("chr1\t10\t20\tGeUtilities_01\t123..45"))
{
// Act
var parser = new BedParser()
{
DropPeakIfInvalidValue = true
};
var parsedData = parser.Parse(file.Path);
// Assert
Assert.False(parsedData.Chromosomes.ContainsKey("chr1"));
}
}
public void BEDWithAllOptionalColumnsAndSomeExtraColumns()
{
// Arrange
var rg = new RegionGenerator {
SummitColumn = 10, StrandColumn = 11
};
using (var file = new TempFileCreator(rg, headerLineCount: 2, peaksCount: 10))
{
// Act
var parser = new BedParser();
var parsedData = parser.Parse(file.Path);
// Assert
Assert.True(parsedData.Chromosomes[rg.Chr].Strands[rg.Strand].Intervals.Count == 10);
}
}
public void MaxLinesToRead(int numberOfPeaksToWrite, uint numberOfPeaksToRead)
{
// Arrange
var rg = new RegionGenerator();
using (var file = new TempFileCreator(rg, peaksCount: numberOfPeaksToWrite))
{
// Act
var parser = new BedParser()
{
MaxLinesToRead = numberOfPeaksToRead
};
var parsedData = parser.Parse(file.Path);
// Assert
Assert.True(parsedData.Chromosomes[rg.Chr].Strands[rg.Strand].Intervals.Count == Math.Min(numberOfPeaksToWrite, numberOfPeaksToRead));
}
}
public void TestStrand(sbyte strandColumn, char strand)
{
// Arrange
var rg = new RegionGenerator
{
Strand = strand,
StrandColumn = strandColumn
};
using (var file = new TempFileCreator(rg))
{
// Act
var parser = new BedParser(rg.Columns);
// Assert
Assert.True(parser.Parse(file.Path).Chromosomes[rg.Chr].Strands.ContainsKey(strand));
}
}
public void ReadNoPeak()
{
// Arrange
var rg = new RegionGenerator();
using (var file = new TempFileCreator(rg, peaksCount: 4))
{
// Act
var parser = new BedParser()
{
MaxLinesToRead = 0
};
var parsedData = parser.Parse(file.Path);
// Assert
Assert.True(parsedData.Chromosomes.Count == 0);
}
}
public void TestSummit(sbyte summitColumn, int summit)
{
// Arrange
var rg = new RegionGenerator {
SummitColumn = summitColumn
};
rg.Summit = summit == -1 ? rg.Left + ((rg.Right - rg.Left) / 2) : summit;
using (var file = new TempFileCreator(rg))
{
// Act
var parser = new BedParser(rg.Columns);
var parsedPeak = parser.Parse(file.Path).Chromosomes[rg.Chr].Strands[rg.Strand].Intervals.ToList()[0];
// Assert
Assert.True(parsedPeak.Summit == rg.Summit);
}
}
public void ThrowExceptionForInvalidCultureInfo()
{
// Arrange
var rg = new RegionGenerator();
using (var file = new TempFileCreator(rg))
{
// Act
var exception = Assert.Throws <ArgumentOutOfRangeException>(
() => new BedParser
{
Culture = "invalid_culture"
});
// Assert
Assert.False(string.IsNullOrEmpty(exception.Message));
Assert.Contains("Invalid culture info", exception.Message);
}
}
public void UseDefaultPValueForPeaksWithInvalidPValue()
{
// Arrange
double defaultValue = 0.112233;
using (var file = new TempFileCreator("chr1\t10\t20\tGeUtilities_01\t123..45"))
{
// Act
var parser = new BedParser()
{
DropPeakIfInvalidValue = false,
DefaultValue = defaultValue
};
var parsedData = parser.Parse(file.Path);
// Assert
Assert.True(parsedData.Chromosomes["chr1"].Strands['*'].Intervals[0].Value == defaultValue);
}
}
public void TestMeanValue()
{
// Arrange
string[] peaks = new string[]
{
"chr1\t10\t20\tGeUtilities_00\t0.1",
"chr1\t30\t40\tGeUtilities_01\t0.01",
"chr3\t50\t60\tGeUtilities_02\t0.001",
"chr4\t70\t80\tGeUtilities_03\t0.0001",
};
using (var file = new TempFileCreator(peaks))
{
// Act
var parser = new BedParser();
var parsedData = parser.Parse(file.Path);
// Assert
Assert.True(parsedData.PValueMean == 0.027775);
}
}
public void PValueConversion(double originalValue, double convertedValue, PValueFormats pvalueFormat)
{
// Arrange
var rg = new RegionGenerator {
Value = convertedValue
};
using (var file = new TempFileCreator(rg))
{
// Act
var parser = new BedParser()
{
PValueFormat = pvalueFormat
};
var parsedData = parser.Parse(file.Path);
// Assert
Assert.True(parsedData.Chromosomes[rg.Chr].Strands[rg.Strand].Intervals[0].Value == originalValue);
}
}
public void DontFailIfGivenColumnNumberIsInvalid()
{
// Arrange
using (var file = new TempFileCreator("chr1\t10\t20"))
{
var parser = new BedParser(new BedColumns()
{
Chr = 0, Left = 1, Right = 2, Name = 3, Strand = 4, Summit = 5, Value = 6
})
{
DropPeakIfInvalidValue = false,
};
// Act
var parsedPeak = parser.Parse(file.Path).Chromosomes["chr1"].Strands['.'].Intervals.ToList()[0];
// Assert
Assert.True(parsedPeak.Left == 10 && parsedPeak.Right == 20);
}
}
public void UseDefaultValueIfInvalidPValueColumn()
{
// Arrange
var rg = new RegionGenerator();
using (var file = new TempFileCreator(rg))
{
// Act
var parser = new BedParser(new BedColumns()
{
Value = 9
})
{
DropPeakIfInvalidValue = false
};
var parsedData = parser.Parse(file.Path);
// Assert
Assert.True(parsedData.Chromosomes.ContainsKey(rg.Chr));
}
}
public void TestMissingChrs()
{
// Arrange
var rg = new RegionGenerator {
Chr = "chr1"
};
using (var file = new TempFileCreator(rg))
{
// Act
var parser = new BedParser()
{
Assembly = Assemblies.hg19,
ReadOnlyAssemblyChrs = false
};
parser.Parse(file.Path);
// Assert
Assert.True(parser.MissingChrs.Count == References.GetGenomeSizes(Assemblies.hg19).Count - 1);
}
}
public void LogErrorIfFailedToReadALine()
{
// Arrange
var rg = new RegionGenerator();
using (var file = new TempFileCreator(rg))
{
// Act
var parser = new BedParser(new BedColumns()
{
Value = 9
})
{
DropPeakIfInvalidValue = true
};
var parsedData = parser.Parse(file.Path);
// Assert
Assert.True(parsedData.Messages.Count > 0);
}
}
public void TestExcessChrs(string chr)
{
// Arrange
var rg = new RegionGenerator {
Chr = chr
};
using (var file = new TempFileCreator(rg))
{
// Act
BedParser <Peak> parser = new BedParser()
{
Assembly = Assemblies.hg19,
ReadOnlyAssemblyChrs = false
};
parser.Parse(file.Path);
// Assert
Assert.True(parser.ExcessChrs.Count == 1);
}
}
public void DropLineIfInvalidChr(string chr)
{
// Arrange
var rg = new RegionGenerator {
Chr = chr
};
using (var file = new TempFileCreator(rg))
{
// Act
var parser = new BedParser()
{
Assembly = Assemblies.hg19,
ReadOnlyAssemblyChrs = true
};
var parsedData = parser.Parse(file.Path);
// Assert
Assert.True(parsedData.Chromosomes.Count == 0);
}
}
public void ParseIntervalChrWhenChrPrefixIsMissing(string chr)
{
// Arrange
var rg = new RegionGenerator {
Chr = chr
};
using (var file = new TempFileCreator(rg))
{
// Act
var parser = new BedParser()
{
Assembly = Assemblies.hg19,
ReadOnlyAssemblyChrs = true
};
var parsedData = parser.Parse(file.Path);
// Assert
Assert.True(parsedData.Chromosomes.ContainsKey("chr" + chr));
}
}
