public void ParseGenomeAssembly_IncorrectFormat_ThrowException()
{
using (var custParser = new CustomAnnotationsParser(GetReadStream("#assembly-GRCh38"), SequenceProvider))
{
Assert.Throws <UserErrorException>(() => custParser.ParseGenomeAssembly());
}
}
public void ReadlineAndCheckPrefix_InvalidPrefix_ThrowException()
{
using (var custParser = new CustomAnnotationsParser(GetReadStream("invalidPrefix=someValue"), SequenceProvider))
{
Assert.Throws <UserErrorException>(() => custParser.ReadlineAndCheckPrefix("expectedPrefix", "anyRow"));
}
}
public void ParseTitle_IncorrectFormat()
{
using (var custParser = new CustomAnnotationsParser(GetReadStream("#title:customSA"), SequenceProvider))
{
Assert.Throws <UserErrorException>(() => custParser.ParseTitle());
}
}
public void ParseGenomeAssembly_UnsupportedAssembly_ThrowException()
{
using (var custParser = new CustomAnnotationsParser(GetReadStream("#assembly=hg19"), SequenceProvider))
{
Assert.Throws <UserErrorException>(() => custParser.ParseGenomeAssembly());
}
}
public void ParseTitle_Conflict_JsonTag()
{
using (var custParser = new CustomAnnotationsParser(GetReadStream("#title=topmed"), SequenceProvider))
{
Assert.Throws <UserErrorException>(() => custParser.ParseTitle());
}
}
public void ParseTags_LessThanFourColumn_ThrowException()
{
const string tagLine = "#CHROM\tPOS\tREF\n";
using (var caParser = new CustomAnnotationsParser(GetReadStream(tagLine), null))
{
Assert.Throws <UserErrorException>(() => caParser.ParseTags());
}
}
public void ParseTypes_InvalidType_ThrowException(string type)
{
string tagAndTypeLines = "#CHROM\tPOS\tREF\tALT\tValue\n" +
$"#type\t.\t.\t.\t{type}";
using (var caParser = new CustomAnnotationsParser(GetReadStream(tagAndTypeLines), null))
{
caParser.ParseTags();
Assert.Throws <UserErrorException>(() => caParser.ParseTypes());
}
}
public void ParseTypes_ValidType_Pass(string type)
{
string tagAndTypeLines = "#CHROM\tPOS\tREF\tALT\tValue\n" +
$"#type\t.\t.\t.\t{type}";
using (var caParser = new CustomAnnotationsParser(GetReadStream(tagAndTypeLines), null))
{
caParser.ParseTags();
caParser.ParseTypes();
}
}
public void ParseHeaderLines_InconsistentFields()
{
const string invalidHeaderLines = "#title=IcslAlleleFrequencies\n" +
"#assembly=GRCh38\n" +
"#CHROM\tPOS\tREF\tALT\tEND\tallAc\tallAn\tallAf\tfailedFilter\tpathogenicity\n" +
"#categories\t.\t.\t.\t.\tAlleleCount\tAlleleNumber\tAlleleFrequency\t.\tPrediction\t.\tMore\n" +
"#descriptions\t.\t.\t.\t.\tALL\tALL\tALL\n" +
"#type\t.\t.\t.\t.\tnumber\tnumber\tnumber\tbool\tstring\tstring";
using (var custParser = new CustomAnnotationsParser(GetReadStream(invalidHeaderLines), null))
{
Assert.Throws <UserErrorException>(() => custParser.ParseHeaderLines());
}
}
public void CheckPosAndRefColumns_InvalidPosOrRef_ThrowException()
{
const string tagLine = "#CHROM\t\tREF\tALT\n";
using (var caParser = new CustomAnnotationsParser(GetReadStream(tagLine), null))
{
Assert.Throws <UserErrorException>(() => caParser.ParseTags());
}
const string tagLine2 = "#CHROM\tPOS\tREFERENCE\tALT\n";
using (var caParser = new CustomAnnotationsParser(GetReadStream(tagLine2), null))
{
Assert.Throws <UserErrorException>(() => caParser.ParseTags());
}
}
public void GetItems_ExtractCustomFilters_failsOnLargeText()
{
const string text = "#title=IcslAlleleFrequencies\n" +
"#assembly=GRCh38\n" +
"#matchVariantsBy=allele\n" +
"#CHROM\tPOS\tREF\tALT\tEND\tallAc\tallAn\tallAf\tfailedFilter\tpathogenicity\tnotes\tanyNumber\tcustomFilter\n" +
"#categories\t.\t.\t.\t.\tAlleleCount\tAlleleNumber\tAlleleFrequency\t.\tPrediction\t.\t.\tFilter\n" +
"#descriptions\t.\t.\t.\t.\tALL\tALL\tALL\t.\t.\t.\t.\t.\n" +
"#type\t.\t.\t.\t.\tnumber\tnumber\tnumber\tbool\tstring\tstring\tnumber\tstring\n" +
"chr1\t12783\tG\tA\t.\t20\t125568\t0.000159\ttrue\tVUS\t\t1.000\tthe good variant, the bad variant and the ugly variant\n";
using (var custParser = CustomAnnotationsParser.Create(GetReadStream(text), SequenceProvider))
{
Assert.Throws <UserErrorException>(() => custParser.GetItems().ToArray());
}
}
public void ExtractItems_TrimmedAndLeftShifted()
{
const string text = "#title=IcslAlleleFrequencies\n" +
"#assembly=GRCh38\n" +
"#matchVariantsBy=allele\n" +
"#CHROM\tPOS\tREF\tALT\tEND\tallAc\tallAn\tallAf\tfailedFilter\tpathogenicity\tnotes\n" +
"#categories\t.\t.\t.\t.\tAlleleCount\tAlleleNumber\tAlleleFrequency\t.\tPrediction\t.\n" +
"#descriptions\t.\t.\t.\t.\tALL\tALL\tALL\t.\t.\t.\n" +
"#type\t.\t.\t.\t.\tnumber\tnumber\tnumber\tbool\tstring\tstring\n";
using (var parser = CustomAnnotationsParser.Create(GetReadStream(text), SequenceProvider))
{
var item = parser.ExtractItems("chr1\t12783\tA\tATA\t.\t20\t125568\t0.000159\ttrue\tVUS\t");
Assert.Equal(12782, item.Position);
Assert.Equal("", item.RefAllele);
Assert.Equal("TA", item.AltAllele);
}
}
public void ParseHeaderLines_AsExpected()
{
const string headerLines = "#title=IcslAlleleFrequencies\n" +
"#assembly=GRCh38\n" +
"#matchVariantsBy=allele\n" +
"#CHROM\tPOS\tREF\tALT\tEND\tallAc\tallAn\tallAf\tfailedFilter\tpathogenicity\tnotes\n" +
"#categories\t.\t.\t.\t.\tAlleleCount\tAlleleNumber\tAlleleFrequency\t.\tPrediction\t.\n" +
"#descriptions\t.\t.\t.\t.\tALL\tALL\tALL\t.\t.\t.\n" +
"#type\t.\t.\t.\t.\tnumber\tnumber\tnumber\tbool\tstring\tstring";
using (var custParser = new CustomAnnotationsParser(GetReadStream(headerLines), null))
{
custParser.ParseHeaderLines();
var expectedJsonKeys = new[]
{ "refAllele", "altAllele", "allAc", "allAn", "allAf", "failedFilter", "pathogenicity", "notes" };
var expectedIntervalJsonKeys = new[]
{ "start", "end", "allAc", "allAn", "allAf", "failedFilter", "pathogenicity", "notes" };
var expectedCategories = new[]
{
CustomAnnotationCategories.AlleleCount, CustomAnnotationCategories.AlleleNumber,
CustomAnnotationCategories.AlleleFrequency, CustomAnnotationCategories.Unknown,
CustomAnnotationCategories.Prediction, CustomAnnotationCategories.Unknown
};
var expectedDescriptions = new[] { "ALL", "ALL", "ALL", null, null, null };
var expectedTypes = new[]
{
SaJsonValueType.Number,
SaJsonValueType.Number,
SaJsonValueType.Number,
SaJsonValueType.Bool,
SaJsonValueType.String,
SaJsonValueType.String
};
Assert.Equal("IcslAlleleFrequencies", custParser.JsonTag);
Assert.Equal(GenomeAssembly.GRCh38, custParser.Assembly);
Assert.True(expectedJsonKeys.SequenceEqual(custParser.JsonKeys));
Assert.True(expectedIntervalJsonKeys.SequenceEqual(custParser.IntervalJsonKeys));
Assert.True(expectedCategories.SequenceEqual(custParser.Categories));
Assert.True(expectedDescriptions.SequenceEqual(custParser.Descriptions));
Assert.Equal(expectedTypes, custParser.ValueTypes);
}
}
public void GetItems_UnsortedData_ThrowException()
{
const string text = "#title=IcslAlleleFrequencies\n" +
"#assembly=GRCh38\n" +
"#matchVariantsBy=allele\n" +
"#CHROM\tPOS\tREF\tALT\tEND\tallAc\tallAn\tallAf\tfailedFilter\tpathogenicity\tnotes\tanyNumber\n" +
"#categories\t.\t.\t.\t.\tAlleleCount\tAlleleNumber\tAlleleFrequency\t.\tPrediction\t.\t.\n" +
"#descriptions\t.\t.\t.\t.\tALL\tALL\tALL\t.\t.\t.\t.\n" +
"#type\t.\t.\t.\t.\tnumber\tnumber\tnumber\tbool\tstring\tstring\tnumber\n" +
"chr1\t12783\tG\tA\t.\t20\t125568\t0.000159\ttrue\tVUS\t\t1.000\n" +
"chr1\t3302\tC\tA\t.\t53\t8928\t0.001421\tfalse\t.\t\t3\n" +
"chr1\t18972\tT\tC\t.\t10\t1000\t0.01\tfalse\t.\t\t100.1234567\n" +
"chr1\t46993\tA\t<DEL>\t50879\t50\t250\t0.001\tfalse\tbenign\t\t3.1415926";
using (var caParser = CustomAnnotationsParser.Create(GetReadStream(text), SequenceProvider))
{
Assert.Throws <UserErrorException>(() => caParser.GetItems().ToArray());
}
}
public void GetItems()
{
const string text = "#title=IcslAlleleFrequencies\n" +
"#assembly=GRCh38\n" +
"#matchVariantsBy=allele\n" +
"#CHROM\tPOS\tREF\tALT\tEND\tallAc\tallAn\tallAf\tfailedFilter\tpathogenicity\tnotes\n" +
"#categories\t.\t.\t.\t.\tAlleleCount\tAlleleNumber\tAlleleFrequency\t.\tPrediction\t.\n" +
"#descriptions\t.\t.\t.\t.\tALL\tALL\tALL\t.\t.\t.\n" +
"#type\t.\t.\t.\t.\tnumber\tnumber\tnumber\tbool\tstring\tstring\n" +
"chr1\t14783\tG\tA\t.\t20\t125568\t0.000159\ttrue\tVUS\t\n" +
"chr2\t10302\tC\tA\t.\t53\t8928\t0.001421\tfalse\t.\t\n" +
"chr2\t46993\tA\t<DEL>\t50879\t50\t250\t0.001\tfalse\tbenign\t";
using (var custParser = CustomAnnotationsParser.Create(GetReadStream(text), SequenceProvider))
{
var items = custParser.GetItems().ToArray();
Assert.Equal(2, items.Length);
Assert.Equal("\"refAllele\":\"G\",\"altAllele\":\"A\",\"allAc\":20,\"allAn\":125568,\"allAf\":0.000159,\"failedFilter\":true,\"pathogenicity\":\"VUS\"", items[0].GetJsonString());
Assert.Equal("\"refAllele\":\"C\",\"altAllele\":\"A\",\"allAc\":53,\"allAn\":8928,\"allAf\":0.001421", items[1].GetJsonString());
}
}
public void IsValidNucleotideSequence_IsValidSequence_Pass()
{
Assert.True(CustomAnnotationsParser.IsValidNucleotideSequence("actgnACTGN"));
Assert.False(CustomAnnotationsParser.IsValidNucleotideSequence("AC-GT"));
}
