public void GenBankFormatterValidateReadAndWriteMultipleDBLinks()
{
// Create a Sequence with all attributes.
// parse and update the properties instead of parsing entire file.
string tempFileName = Path.GetTempFileName();
ISequenceParser parser1 = new GenBankParser();
using (parser1.Open(_genBankFile_WithMultipleDBLines))
{
var orgSeq = parser1.Parse().First();
ISequenceFormatter formatter = new GenBankFormatter();
using (formatter.Open(tempFileName))
{
formatter.Format(orgSeq);
formatter.Close();
}
}
var same = CompareFiles(tempFileName, _genBankFile_WithMultipleDBLines);
File.Delete(tempFileName);
Assert.IsTrue(same);
ApplicationLog.WriteLine("GenBank Formatter: Successful read->write loop");
}
public void TestGenBankWhenUserSetsProteinAlphabet()
{
// set correct alphabet and parse
ISequenceParser parser = new GenBankParser();
parser.Alphabet = Alphabets.DNA;
ISequence seq = parser.Parse(_singleProteinSeqGenBankFilename).FirstOrDefault();
Assert.AreEqual(Alphabets.DNA, seq.Alphabet);
// format
ISequenceFormatter formatter = new GenBankFormatter();
using (formatter.Open(TempGenBankFileName))
formatter.Format(seq);
string actual = string.Empty;
using (StreamReader reader = new StreamReader(TempGenBankFileName))
{
actual = reader.ReadToEnd();
}
File.Delete(TempGenBankFileName);
// test the formatting
Assert.AreEqual(_singleProteinSeqGenBankFileExpectedOutput.Replace(" ", "").Replace("\r\n", Environment.NewLine), actual.Replace(" ", ""));
}
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);
IEnumerable<ISequence> seqList = new GenBankParser().Parse(fileInfo.FullName);
ISequenceFormatter formatter = new GenBankFormatter();
using (formatter.Open(TempGenBankFileName))
{
(formatter as GenBankFormatter).Format(seqList.ToList());
}
using (var reader = new StreamReader(TempGenBankFileName))
{
string actual = reader.ReadToEnd();
}
File.Delete(TempGenBankFileName);
}
}
public void GenBankFormatterValidateWriteUsingStream()
{
InitializeXmlVariables();
// Create a Sequence with all attributes.
// Parse and update the properties instead of parsing entire file.
ISequenceParser parser1 = new GenBankParser();
{
IEnumerable<ISequence> seqList1 = parser1.Parse(FilePath);
string tempFileName = Path.GetTempFileName();
GenBankMetadata metadata = null;
ISequence seq = null;
string expectedUpdatedSequence =
ExpectedSequence.Replace("\r", "").Replace("\n", "").Replace(" ", "");
var orgSeq =
new Sequence(Utility.GetAlphabet(AlphabetName), expectedUpdatedSequence);
orgSeq.Metadata.Add("GenBank",
seqList1.ElementAt(0).Metadata["GenBank"]);
orgSeq.ID = seqList1.ElementAt(0).ID;
ISequenceFormatter formatter = new GenBankFormatter();
{
using (formatter.Open(tempFileName))
{
formatter.Format(orgSeq);
}
}
var parserObj = new GenBankParser();
{
IEnumerable<ISequence> seqList = parserObj.Parse(tempFileName);
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
metadata = (GenBankMetadata) seq.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(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");
File.Delete(tempFileName);
}
}