public void GenBankFormatterValidateWriteWithFilePath()
{
InitializeXmlVariables();
using (ISequenceParser parserObj = new GenBankParser(FilePath))
{
IEnumerable <ISequence> seqList1 = parserObj.Parse();
string tempFileName = System.IO.Path.GetTempFileName();
string expectedUpdatedSequence =
ExpectedSequence.Replace("\r", "").Replace("\n", "").Replace(" ", "");
Sequence orgSeq = new Sequence(Utility.GetAlphabet(AlphabetName), expectedUpdatedSequence);
orgSeq.ID = seqList1.ElementAt(0).ID;
orgSeq.Metadata.Add("GenBank", (GenBankMetadata)seqList1.ElementAt(0).Metadata["GenBank"]);
using (ISequenceFormatter formatter = new GenBankFormatter(tempFileName))
{
formatter.Write(orgSeq);
formatter.Close();
// parse
ISequenceParser parserObjFromFile = new GenBankParser(tempFileName);
IEnumerable <ISequence> seqList =
parserObjFromFile.Parse();
ISequence seq = seqList.ElementAt(0);
Assert.AreEqual(Utility.GetAlphabet(AlphabetName), seq.Alphabet);
Assert.AreEqual(SeqId, seq.ID);
ApplicationLog.WriteLine(
"GenBank Formatter BVT: Successfully validated the Alphabet, Molecular type, Sequence ID and Display ID");
// test the metadata that is tricky to parse, and will not be tested implicitly by
// testing the formatting
GenBankMetadata metadata =
(GenBankMetadata)orgSeq.Metadata["GenBank"];
if (metadata.Locus.Strand != SequenceStrandType.None)
{
Assert.AreEqual(StrandType,
metadata.Locus.Strand.ToString());
}
Assert.AreEqual(StrandTopology.ToUpper(CultureInfo.CurrentCulture),
metadata.Locus.StrandTopology.ToString().ToUpper(CultureInfo.CurrentCulture));
Assert.AreEqual(Div, metadata.Locus.DivisionCode.ToString());
Assert.AreEqual(DateTime.Parse(SequenceDate, null),
metadata.Locus.Date);
Assert.AreEqual(Version, metadata.Version.Version.ToString((IFormatProvider)null));
Assert.AreEqual(PrimaryId, metadata.Version.GiNumber);
ApplicationLog.WriteLine(
"GenBank Formatter BVT: Successfully validated the StrandType, StrandTopology, Division, Date, Version, PrimaryID Properties");
// test the sequence string
Assert.AreEqual(ExpectedSequence, new string(seq.Select(a => (char)a).ToArray()));
ApplicationLog.WriteLine(
"GenBank Formatter BVT: Successfully validated the Sequence");
Console.WriteLine(string.Format((IFormatProvider)null,
"GenBank Formatter BVT: Successfully validated the Sequence '{0}'",
ExpectedSequence));
parserObjFromFile.Close();
parserObjFromFile.Dispose();
File.Delete(tempFileName);
}
}
}
public void GenBankFeatures()
{
// parse
ISequenceParser parser = new GenBankParser(_singleProteinSeqGenBankFilename);
ISequence seq = parser.Parse().FirstOrDefault();
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);
parser.Close();
parser.Dispose();
ISequenceParser parser1 = new GenBankParser(_genBankDataPath + @"\NC_001284.gbk");
ISequence seq1 = parser1.Parse().FirstOrDefault();
metadata = seq1.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);
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 = ASCIIEncoding.ASCII.GetString(tempData);
Assert.AreEqual(metadata.Features.CodingSequences[0].Translation.Trim('"'), sequenceInString.Trim('"'));
Assert.AreEqual(2, metadata.GetFeatures(11918, 12241).Count);
}
public void TestGenBankForManyFiles()
{
// parser and formatter will be used for all files in input dir
// iterate through the files in input dir, parsing and formatting each; write results
// to log file
DirectoryInfo inputDirInfo = new DirectoryInfo(_genBankDataPath);
foreach (FileInfo fileInfo in inputDirInfo.GetFiles("*.gbk"))
{
ApplicationLog.WriteLine("Parsing file {0}...{1}", fileInfo.FullName, Environment.NewLine);
ISequenceParser parser = new GenBankParser(fileInfo.FullName);
try
{
IEnumerable <ISequence> seqList = parser.Parse();
// don't do anything with it; just make sure it doesn't crash
ISequenceFormatter formatter = new GenBankFormatter(TempGenBankFileName);
string actual = string.Empty;
(formatter as GenBankFormatter).Write(seqList.ToList());
using (StreamReader reader = new StreamReader(TempGenBankFileName))
{
actual = reader.ReadToEnd();
}
File.Delete(TempGenBankFileName);
parser.Close();
parser.Dispose();
ApplicationLog.WriteLine("Parse completed successfully." + Environment.NewLine);
}
catch (Exception e)
{
ApplicationLog.WriteLine("Parse failed:" + Environment.NewLine + e + Environment.NewLine);
throw;
}
}
}