public void GenBankFeatures()
{
// parse
ISequenceParser parser = new GenBankParser();
ISequence seq = parser.ParseOne(_singleProteinSeqGenBankFilename);
GenBankMetadata metadata = seq.Metadata["GenBank"] as GenBankMetadata;
List <CodingSequence> CDS = metadata.Features.CodingSequences;
Assert.AreEqual(CDS.Count, 3);
Assert.AreEqual(CDS[0].DatabaseCrossReference.Count, 1);
Assert.AreEqual(CDS[0].GeneSymbol, string.Empty);
Assert.AreEqual(metadata.Features.GetFeatures("source").Count, 1);
Assert.IsFalse(CDS[0].Pseudo);
Assert.AreEqual(metadata.GetFeatures(1, 109).Count, 2);
Assert.AreEqual(metadata.GetFeatures(1, 10).Count, 2);
Assert.AreEqual(metadata.GetFeatures(10, 100).Count, 2);
Assert.AreEqual(metadata.GetFeatures(120, 150).Count, 2);
Assert.AreEqual(metadata.GetCitationsReferredInFeatures().Count, 0);
seq = parser.ParseOne(_genBankDataPath + @"\NC_001284.gbk");
metadata = seq.Metadata["GenBank"] as GenBankMetadata;
Assert.AreEqual(metadata.Features.All.Count, 743);
Assert.AreEqual(metadata.Features.CodingSequences.Count, 117);
Assert.AreEqual(metadata.Features.Exons.Count, 32);
Assert.AreEqual(metadata.Features.Introns.Count, 22);
Assert.AreEqual(metadata.Features.Genes.Count, 60);
Assert.AreEqual(metadata.Features.MiscFeatures.Count, 455);
Assert.AreEqual(metadata.Features.Promoters.Count, 17);
Assert.AreEqual(metadata.Features.TransferRNAs.Count, 21);
Assert.AreEqual(metadata.Features.All.FindAll(F => F.Key.Equals(StandardFeatureKeys.CodingSequence)).Count, 117);
Assert.AreEqual(metadata.Features.GetFeatures(StandardFeatureKeys.CodingSequence).Count, 117);
Assert.AreEqual(metadata.Features.CodingSequences[0].Translation.Trim('"'), metadata.Features.CodingSequences[0].GetTranslation().ToString());
Assert.AreEqual(2, metadata.GetFeatures(11918, 12241).Count);
}
public void GenBankFeatures()
{
// parse
ISequence seq = new GenBankParser()
.Parse(_singleProteinSeqGenBankFilename)
.FirstOrDefault();
Assert.IsNotNull(seq);
GenBankMetadata metadata = seq.Metadata["GenBank"] as GenBankMetadata;
Assert.IsNotNull(metadata);
List <CodingSequence> CDS = metadata.Features.CodingSequences;
Assert.AreEqual(CDS.Count, 3);
Assert.AreEqual(CDS[0].DatabaseCrossReference.Count, 1);
Assert.AreEqual(CDS[0].GeneSymbol, string.Empty);
Assert.AreEqual(metadata.Features.GetFeatures("source").Count, 1);
Assert.IsFalse(CDS[0].Pseudo);
Assert.AreEqual(metadata.GetFeatures(1, 109).Count, 2);
Assert.AreEqual(metadata.GetFeatures(1, 10).Count, 2);
Assert.AreEqual(metadata.GetFeatures(10, 100).Count, 2);
Assert.AreEqual(metadata.GetFeatures(120, 150).Count, 2);
Assert.AreEqual(metadata.GetCitationsReferredInFeatures().Count, 0);
ISequence seq1 = new GenBankParser()
.Parse(Path.Combine(_genBankDataPath, "NC_001284.gbk"))
.FirstOrDefault();
Assert.IsNotNull(seq1);
metadata = seq1.Metadata["GenBank"] as GenBankMetadata;
Assert.IsNotNull(metadata);
Assert.AreEqual(metadata.Features.All.Count, 743);
Assert.AreEqual(metadata.Features.CodingSequences.Count, 117);
Assert.AreEqual(metadata.Features.Exons.Count, 32);
Assert.AreEqual(metadata.Features.Introns.Count, 22);
Assert.AreEqual(metadata.Features.Genes.Count, 60);
Assert.AreEqual(metadata.Features.MiscFeatures.Count, 455);
Assert.AreEqual(metadata.Features.Promoters.Count, 17);
Assert.AreEqual(metadata.Features.TransferRNAs.Count, 21);
Assert.AreEqual(metadata.Features.All.FindAll(F => F.Key.Equals(StandardFeatureKeys.CodingSequence)).Count, 117);
Assert.AreEqual(metadata.Features.GetFeatures(StandardFeatureKeys.CodingSequence).Count, 117);
ISequence seqTemp = metadata.Features.CodingSequences[0].GetTranslation();
byte[] tempData = new byte[seqTemp.Count];
for (int i = 0; i < seqTemp.Count; i++)
{
tempData[i] = seqTemp[i];
}
string sequenceInString = Encoding.ASCII.GetString(tempData);
Assert.AreEqual(metadata.Features.CodingSequences[0].Translation.Trim('"'), sequenceInString.Trim('"'));
Assert.AreEqual(2, metadata.GetFeatures(11918, 12241).Count);
}