private void AddRegulatoryRegion(IAnnotatedVariant annotatedVariant)
{
var regulatoryRegion = new RegulatoryRegion(_chromosome, 103, 104, CompactId.Convert("7157"),
RegulatoryRegionType.TF_binding_site);
var consequences = new List <ConsequenceTag> {
ConsequenceTag.regulatory_region_amplification
};
annotatedVariant.RegulatoryRegions.Add(new AnnotatedRegulatoryRegion(regulatoryRegion, consequences));
}
/// <summary>
/// parses the database cache file and populates the specified lists and interval trees
/// </summary>
public TranscriptCacheData Read(IDictionary <ushort, IChromosome> refIndexToChromosome)
{
var genes = ReadItems(_reader, () => Gene.Read(_reader, refIndexToChromosome));
var transcriptRegions = ReadItems(_reader, () => TranscriptRegion.Read(_reader));
var mirnas = ReadItems(_reader, () => Interval.Read(_reader));
var peptideSeqs = ReadItems(_reader, () => _reader.ReadAsciiString());
var regulatoryRegions = ReadIntervals(_reader, () => RegulatoryRegion.Read(_reader, refIndexToChromosome));
var transcripts = ReadIntervals(_reader, () => Transcript.Read(_reader, refIndexToChromosome, genes, transcriptRegions, mirnas, peptideSeqs));
return(new TranscriptCacheData(Header, genes, transcriptRegions, mirnas, peptideSeqs, transcripts, regulatoryRegions));
}
private IRegulatoryRegion[] GetRegulatoryRegions()
{
var regulatoryRegions = new IRegulatoryRegion[7];
regulatoryRegions[0] = new RegulatoryRegion(_chr17, 100,200,CompactId.Empty, RegulatoryRegionType.enhancer);
regulatoryRegions[1] = new RegulatoryRegion(_chr17, 200, 300, CompactId.Empty, RegulatoryRegionType.enhancer);
regulatoryRegions[2] = new RegulatoryRegion(_chr17, 300, 400, CompactId.Empty, RegulatoryRegionType.enhancer);
regulatoryRegions[3] = new RegulatoryRegion(_chr1, 1000, 2000, CompactId.Empty, RegulatoryRegionType.enhancer);
regulatoryRegions[4] = new RegulatoryRegion(_chr1, 2000, 3000, CompactId.Empty, RegulatoryRegionType.enhancer);
regulatoryRegions[5] = new RegulatoryRegion(_chr11, 150, 250, CompactId.Empty, RegulatoryRegionType.enhancer);
regulatoryRegions[6] = new RegulatoryRegion(_chr11, 250, 350, CompactId.Empty, RegulatoryRegionType.enhancer);
return regulatoryRegions;
}
public TranscriptCacheReaderTests()
{
var chr1 = new Chromosome("chr1", "1", 0);
var chr2 = new Chromosome("chr2", "2", 1);
var chr3 = new Chromosome("chr3", "3", 2);
_refIndexToChromosome = new Dictionary <ushort, IChromosome>
{
[chr1.Index] = chr1,
[chr2.Index] = chr2,
[chr3.Index] = chr3
};
const GenomeAssembly genomeAssembly = GenomeAssembly.GRCh38;
var baseHeader = new Header("test", 2, 3, Source.BothRefSeqAndEnsembl, 4, genomeAssembly);
var customHeader = new TranscriptCacheCustomHeader(1, 2);
_expectedHeader = new CacheHeader(baseHeader, customHeader);
var transcriptRegions = new ITranscriptRegion[]
{
new TranscriptRegion(TranscriptRegionType.Exon, 1, 100, 199, 300, 399),
new TranscriptRegion(TranscriptRegionType.Intron, 1, 200, 299, 399, 400),
new TranscriptRegion(TranscriptRegionType.Exon, 2, 300, 399, 400, 499)
};
var mirnas = new IInterval[2];
mirnas[0] = new Interval(100, 200);
mirnas[1] = new Interval(300, 400);
var peptideSeqs = new[] { "MASE*" };
var genes = new IGene[1];
genes[0] = new Gene(chr3, 100, 200, true, "TP53", 300, CompactId.Convert("7157"),
CompactId.Convert("ENSG00000141510"));
var regulatoryRegions = new IRegulatoryRegion[2];
regulatoryRegions[0] = new RegulatoryRegion(chr3, 1200, 1300, CompactId.Convert("123"), RegulatoryRegionType.enhancer);
regulatoryRegions[1] = new RegulatoryRegion(chr3, 1250, 1450, CompactId.Convert("456"), RegulatoryRegionType.enhancer);
var regulatoryRegionIntervalArrays = regulatoryRegions.ToIntervalArrays(3);
var transcripts = GetTranscripts(chr3, genes, transcriptRegions, mirnas);
var transcriptIntervalArrays = transcripts.ToIntervalArrays(3);
_expectedCacheData = new TranscriptCacheData(_expectedHeader, genes, transcriptRegions, mirnas, peptideSeqs,
transcriptIntervalArrays, regulatoryRegionIntervalArrays);
}
private static Stream GetCacheStream()
{
const GenomeAssembly genomeAssembly = GenomeAssembly.GRCh38;
var baseHeader = new Header("test", 2, 3, Source.BothRefSeqAndEnsembl, 4, genomeAssembly);
var customHeader = new TranscriptCacheCustomHeader(1, 2);
var expectedHeader = new CacheHeader(baseHeader, customHeader);
var transcriptRegions = new ITranscriptRegion[]
{
new TranscriptRegion(TranscriptRegionType.Exon, 1, 100, 199, 300, 399),
new TranscriptRegion(TranscriptRegionType.Intron, 1, 200, 299, 399, 400),
new TranscriptRegion(TranscriptRegionType.Exon, 2, 300, 399, 400, 499)
};
var mirnas = new IInterval[2];
mirnas[0] = new Interval(100, 200);
mirnas[1] = new Interval(300, 400);
var peptideSeqs = new[] { "MASE*" };
var genes = new IGene[1];
genes[0] = new Gene(ChromosomeUtilities.Chr3, 100, 200, true, "TP53", 300, CompactId.Convert("7157"),
CompactId.Convert("ENSG00000141510"));
var regulatoryRegions = new IRegulatoryRegion[2];
regulatoryRegions[0] = new RegulatoryRegion(ChromosomeUtilities.Chr3, 1200, 1300, CompactId.Convert("123"), RegulatoryRegionType.enhancer);
regulatoryRegions[1] = new RegulatoryRegion(ChromosomeUtilities.Chr3, 1250, 1450, CompactId.Convert("456"), RegulatoryRegionType.enhancer);
var regulatoryRegionIntervalArrays = regulatoryRegions.ToIntervalArrays(3);
var transcripts = GetTranscripts(ChromosomeUtilities.Chr3, genes, transcriptRegions, mirnas);
var transcriptIntervalArrays = transcripts.ToIntervalArrays(3);
var expectedCacheData = new TranscriptCacheData(expectedHeader, genes, transcriptRegions, mirnas, peptideSeqs,
transcriptIntervalArrays, regulatoryRegionIntervalArrays);
var ms = new MemoryStream();
using (var writer = new TranscriptCacheWriter(ms, expectedHeader, true))
{
writer.Write(expectedCacheData);
}
ms.Position = 0;
return(ms);
}
private IRegulatoryRegion[] GetRegulatoryRegions()
{
var regulatoryRegions = new IRegulatoryRegion[3];
regulatoryRegions[0] = new RegulatoryRegion(_chr11, 11000, 12000, CompactId.Empty,
RegulatoryRegionType.promoter);
regulatoryRegions[1] = new RegulatoryRegion(_chr1, 120, 180, CompactId.Empty,
RegulatoryRegionType.promoter);
regulatoryRegions[2] = new RegulatoryRegion(_chr1, 300, 320, CompactId.Empty,
RegulatoryRegionType.promoter);
return(regulatoryRegions);
}
private static IRegulatoryRegion[] GetRegulatoryRegions()
{
var regulatoryRegions = new IRegulatoryRegion[3];
regulatoryRegions[0] = new RegulatoryRegion(ChromosomeUtilities.Chr11, 11000, 12000, CompactId.Empty,
RegulatoryRegionType.promoter);
regulatoryRegions[1] = new RegulatoryRegion(ChromosomeUtilities.Chr1, 120, 180, CompactId.Empty,
RegulatoryRegionType.promoter);
regulatoryRegions[2] = new RegulatoryRegion(ChromosomeUtilities.Chr1, 300, 320, CompactId.Empty,
RegulatoryRegionType.promoter);
return(regulatoryRegions);
}
public void RegulatoryRegion_EndToEnd()
{
IChromosome expectedChromosome = new Chromosome("chrBob", "Bob", 1);
const int expectedStart = int.MaxValue;
const int expectedEnd = int.MinValue;
const string expectedId = "ENST00000540021";
const RegulatoryRegionType expectedType = RegulatoryRegionType.open_chromatin_region;
var indexToChromosome = new Dictionary <ushort, IChromosome>
{
[expectedChromosome.Index] = expectedChromosome
};
// ReSharper disable ConditionIsAlwaysTrueOrFalse
var regulatoryRegion = new RegulatoryRegion(expectedChromosome, expectedStart, expectedEnd,
CompactId.Convert(expectedId), expectedType);
// ReSharper restore ConditionIsAlwaysTrueOrFalse
IRegulatoryRegion observedRegulatoryRegion;
using (var ms = new MemoryStream())
{
using (var writer = new ExtendedBinaryWriter(ms, Encoding.UTF8, true))
{
regulatoryRegion.Write(writer);
}
ms.Position = 0;
using (var reader = new ExtendedBinaryReader(ms))
{
observedRegulatoryRegion = RegulatoryRegion.Read(reader, indexToChromosome);
}
}
Assert.NotNull(observedRegulatoryRegion);
Assert.Equal(expectedStart, observedRegulatoryRegion.Start);
Assert.Equal(expectedEnd, observedRegulatoryRegion.End);
Assert.Equal(expectedId, observedRegulatoryRegion.Id.WithoutVersion);
Assert.Equal(expectedType, observedRegulatoryRegion.Type);
Assert.Equal(expectedChromosome.Index, observedRegulatoryRegion.Chromosome.Index);
}