public void ValidateBasicDerivedSequenceWithFastaFormat()
{
// Gets the expected sequence from the Xml
string expectedSequence = _utilityObj._xmlUtil.GetTextValue(
Constants.SimpleFastaNodeName, Constants.ExpectedSequenceNode);
string fastAFilePath = _utilityObj._xmlUtil.GetTextValue(
Constants.SimpleFastaNodeName, Constants.FilePathNode);
Assert.IsTrue(File.Exists(fastAFilePath));
// Logs information to the log file
ApplicationLog.WriteLine(string.Concat(
"Sequence BVT: The File exist in the Path ", fastAFilePath));
IList <ISequence> sequence = null;
using (FastaParser parser = new FastaParser())
{
//Parse a FastA file Using Parse method and convert the same to sequence.
sequence = parser.Parse(fastAFilePath);
}
Assert.IsNotNull(sequence);
Sequence fastASequence = (Sequence)sequence[0];
Assert.IsNotNull(fastASequence);
Assert.AreEqual(expectedSequence, fastASequence.ToString());
ApplicationLog.WriteLine(string.Concat(
"Sequence BVT: The Sequence is as expected."));
byte[] tmpEncodedSeq = new byte[fastASequence.Count];
(fastASequence as IList <byte>).CopyTo(tmpEncodedSeq, 0);
Assert.AreEqual(expectedSequence.Length, tmpEncodedSeq.Length);
ApplicationLog.WriteLine(string.Concat(
"Sequence BVT: The Sequence Length is as expected."));
Assert.AreEqual(_utilityObj._xmlUtil.GetTextValue(Constants.SimpleProteinAlphabetNode, Constants.SequenceIdNode), fastASequence.ID);
ApplicationLog.WriteLine(string.Concat(
"Sequence BVT: SequenceID is as expected."));
Assert.AreEqual(fastASequence.Alphabet.Name, _utilityObj._xmlUtil.GetTextValue(
Constants.SimpleFastaNodeName, Constants.AlphabetNameNode));
ApplicationLog.WriteLine(string.Concat(
"Sequence BVT: The Sequence Alphabet is as expected."));
// Create a derived Sequences for the fastA file sequence.
BasicDerivedSequence fastADerivedSeq =
new BasicDerivedSequence(fastASequence, false, false, -1, -1);
// validate the DerivedSequence with originalSequence.
Assert.IsNotNull(fastADerivedSeq);
Assert.AreEqual(expectedSequence, fastADerivedSeq.ToString());
Assert.AreEqual(fastASequence.ToString(), fastADerivedSeq.ToString());
ApplicationLog.WriteLine(string.Concat(
"Sequence BVT: The BasicDerived Sequence is as expected."));
// Logs to Nunit GUI.
Console.WriteLine(
"Sequence BVT: Validation of FastaA file Sequence is completed successfully.");
}
public void ValidateBasicDerivedSequence()
{
// Gets the actual sequence and the alphabet from the Xml
string alphabetName = _utilityObj._xmlUtil.GetTextValue(
Constants.SimpleDnaAlphabetNode, Constants.AlphabetNameNode);
string actualSequence = _utilityObj._xmlUtil.GetTextValue(
Constants.SimpleDnaAlphabetNode, Constants.ExpectedNormalString);
// Logs information to the log file
ApplicationLog.WriteLine(string.Format((IFormatProvider)null,
"Sequence BVT: Sequence '{0}' and Alphabet '{1}'.", actualSequence, alphabetName));
Sequence createSequence = new Sequence(
Utility.GetAlphabet(alphabetName), actualSequence);
Assert.IsNotNull(createSequence);
// Validate the createdSequence
Assert.AreEqual(createSequence.ToString(), actualSequence);
ApplicationLog.WriteLine(string.Format((IFormatProvider)null,
"Sequence BVT: Sequence {0} is as expected.", createSequence.ToString()));
// Logs to the NUnit GUI (Console.Out) window
Console.WriteLine(string.Format((IFormatProvider)null,
"Sequence BVT: Sequence {0} is as expected.", createSequence.ToString()));
Assert.AreEqual(Utility.GetAlphabet(alphabetName), createSequence.Alphabet);
ApplicationLog.WriteLine(string.Format((IFormatProvider)null,
"Sequence BVT: Sequence Alphabet is '{0}' and is as expected.",
createSequence.Alphabet.Name));
//Create a BasicDerived Sequence.
BasicDerivedSequence derivedSequence = new BasicDerivedSequence(
createSequence, false, false, -1, -1);
//Validate the DerivedSequence
Assert.AreEqual(createSequence.ToString(), derivedSequence.ToString());
ApplicationLog.WriteLine(string.Format((IFormatProvider)null,
"Sequence BVT: Sequence {0} is as expected.", derivedSequence.ToString()));
Assert.IsNotNull(derivedSequence);
Assert.AreEqual(derivedSequence.ToString(), actualSequence);
Assert.AreEqual(Utility.GetAlphabet(alphabetName), derivedSequence.Alphabet);
ApplicationLog.WriteLine(string.Format((IFormatProvider)null,
"Sequence BVT: Sequence Alphabet is '{0}' and is as expected.",
derivedSequence.Alphabet.Name));
// Logs to the NUnit GUI (Console.Out) window
ApplicationLog.WriteLine(
"Sequence BVT: The BasicDerived Sequence validation is completed successfully.");
}
public void TestBasicDerivedSequenceWithBinaryFormatter()
{
Stream stream = null;
try
{
stream = File.Open("BasicDerivedSequence.data", FileMode.Create);
BinaryFormatter formatter = new BinaryFormatter();
string id = Guid.NewGuid().ToString();
Sequence seq1 = new Sequence(Alphabets.RNA, "ACUGCA");
seq1.ID = id;
seq1.DisplayID = "displayid";
seq1.Documentation = "document";
BasicDerivedSequence seq = new BasicDerivedSequence(seq1, true, true, -1, -1);
formatter.Serialize(stream, seq);
stream.Seek(0, SeekOrigin.Begin);
BasicDerivedSequence deserializedSeq = (BasicDerivedSequence)formatter.Deserialize(stream);
Assert.AreNotSame(seq, deserializedSeq);
Assert.AreSame(seq.Alphabet, deserializedSeq.Alphabet);
Assert.AreEqual(seq.Complement.ToString(), deserializedSeq.Complement.ToString());
Assert.AreEqual(seq.Complemented, deserializedSeq.Complemented);
Assert.AreEqual(seq.Count, deserializedSeq.Count);
Assert.AreEqual(seq.DisplayID, deserializedSeq.DisplayID);
Assert.AreEqual(seq.Documentation, deserializedSeq.Documentation);
Assert.AreEqual(seq.ID, deserializedSeq.ID);
Assert.AreEqual(seq.IsReadOnly, deserializedSeq.IsReadOnly);
Assert.AreEqual(seq.MoleculeType, deserializedSeq.MoleculeType);
Assert.AreEqual(seq.RangeLength, deserializedSeq.RangeLength);
Assert.AreEqual(seq.RangeStart, deserializedSeq.RangeStart);
Assert.AreEqual(seq.Reverse.ToString(), deserializedSeq.Reverse.ToString());
Assert.AreEqual(seq.ReverseComplement.ToString(), deserializedSeq.ReverseComplement.ToString());
Assert.AreEqual(seq.Reversed, deserializedSeq.Reversed);
Assert.AreEqual(seq.Source.ToString(), deserializedSeq.Source.ToString());
Assert.AreEqual(seq.ToString(), deserializedSeq.ToString());
}
catch (Exception)
{
Assert.Fail();
}
finally
{
if (stream != null)
{
stream.Close();
stream = null;
}
}
}
public void ValidateReverse()
{
// Gets the actual sequence and the alphabet from the Xml
string alphabetName = _utilityObj._xmlUtil.GetTextValue(
Constants.SimpleDnaAlphabetNode, Constants.AlphabetNameNode);
string actualSequence = _utilityObj._xmlUtil.GetTextValue(
Constants.SimpleDnaAlphabetNode, Constants.ExpectedNormalString);
string expectedRevSequence = _utilityObj._xmlUtil.GetTextValue(
Constants.SimpleDnaAlphabetNode, Constants.ExpectedReverseSequence);
string expectedDnaNormalSequenceCount = _utilityObj._xmlUtil.GetTextValue(
Constants.SimpleDnaAlphabetNode, Constants.EncodedDnaNormalSequenceCount);
string revDerSeq = string.Empty;
// Logs information to the log file
ApplicationLog.WriteLine(string.Format((IFormatProvider)null,
"Sequence BVT: Sequence '{0}' and Alphabet '{1}'.", actualSequence, alphabetName));
Sequence createSequence = new Sequence(Utility.GetAlphabet(alphabetName), actualSequence);
Assert.IsNotNull(createSequence);
// Validate the createdSequence
Assert.AreEqual(createSequence.ToString(), actualSequence);
ApplicationLog.WriteLine(string.Format((IFormatProvider)null,
"Sequence BVT: Sequence {0} is as expected.", createSequence.ToString()));
// Logs to the NUnit GUI (Console.Out) window
Console.WriteLine(string.Format((IFormatProvider)null,
"Sequence BVT: Sequence {0} is as expected.", createSequence.ToString()));
Assert.AreEqual(Utility.GetAlphabet(alphabetName), createSequence.Alphabet);
ApplicationLog.WriteLine(string.Format((IFormatProvider)null,
"Sequence Alphabet is '{0}' and is as expected.", createSequence.Alphabet.Name));
// Create a BasicDerived Sequence.
BasicDerivedSequence derivedSequence = new BasicDerivedSequence(
createSequence, false, false, -1, -1);
// Validate the Reverse of DerivedSequence.
revDerSeq = derivedSequence.Reverse.ToString();
Assert.AreEqual(revDerSeq.Length.ToString((IFormatProvider)null),
expectedDnaNormalSequenceCount);
Assert.AreEqual(revDerSeq, expectedRevSequence);
ApplicationLog.WriteLine(string.Format((IFormatProvider)null,
"Sequence BVT: Reverse sequence {0} is expected", revDerSeq));
// Logs to the NUnit GUI (Console.Out) window
ApplicationLog.WriteLine(
"Sequence BVT: The Reverse of the sequence is validated successfully.");
}
public void ValidateBasicDerivedSequenceWithGenBankFormat()
{
// Gets the expected sequence from the Xml
string expectedSequence = _utilityObj._xmlUtil.GetTextValue(
Constants.SimpleGeneBankNodeName, Constants.ExpectedSequenceNode);
string geneBankFilePath = _utilityObj._xmlUtil.GetTextValue(
Constants.SimpleGeneBankNodeName, Constants.FilePathNode);
Assert.IsTrue(File.Exists(geneBankFilePath));
// Logs information to the log file
ApplicationLog.WriteLine(string.Concat(
"Sequence BVT: Sequence BVT: The File exist in the Path ", geneBankFilePath));
// Parse a GenBank file Using Parse method and convert the same to sequence.
ISequenceParser parser = new GenBankParser();
IList <ISequence> sequence = parser.Parse(geneBankFilePath);
Assert.IsNotNull(sequence);
Sequence geneBankSeq = (Sequence)sequence[0];
Assert.IsNotNull(geneBankSeq);
Assert.AreEqual(expectedSequence, geneBankSeq.ToString());
ApplicationLog.WriteLine(string.Concat(
"Sequence BVT: The GenBank Sequence is as expected."));
byte[] tmpEncodedSeq = new byte[geneBankSeq.Count];
(geneBankSeq as IList <byte>).CopyTo(tmpEncodedSeq, 0);
Assert.AreEqual(expectedSequence.Length, tmpEncodedSeq.Length);
ApplicationLog.WriteLine(string.Concat(
"Sequence BVT: The GenBank Sequence Length is as expected."));
// Create a derived Sequences for the fastA file sequence.
BasicDerivedSequence genebankDerivedSeq =
new BasicDerivedSequence(geneBankSeq, false, false, -1, -1);
// validate the DerivedSequence with originalSequence.
Assert.IsNotNull(genebankDerivedSeq);
Assert.AreEqual(expectedSequence, genebankDerivedSeq.ToString());
Assert.AreEqual(geneBankSeq.ToString(), genebankDerivedSeq.ToString());
ApplicationLog.WriteLine(string.Concat(
"Sequence BVT: The BasicDerived Sequence is as expected."));
// Logs to Nunit GUI.
Console.WriteLine(
"Sequence BVT: Validation of GenBank file Sequence is completed successfully.");
}
public void ValidateDerivedSequenceFindMatches()
{
ISequence seq = new Sequence(Alphabets.DNA, "ACCGGTT");
BasicDerivedSequence bdsObj =
new BasicDerivedSequence(seq, false, false, 0, seq.Count - 1);
IList <string> patterns = new List <string>();
patterns.Add("ACCG");
IDictionary <string, IList <int> > actual = bdsObj.FindMatches(patterns, 0, false);
Assert.AreEqual(1, actual.Count);
Assert.AreEqual(0, actual["ACCG"][0]);
Console.WriteLine("Derived Sequence : Successfully validated FindMatches() method.");
ApplicationLog.WriteLine("Derived Sequence : Successfully validated FindMatches() method.");
}
public void BasicDerivedSequenceClone()
{
ApplicationLog.WriteLine("BasicDerivedSequenceClone test started");
string seqData = "GCCAACGAACCGGAAACCCGGGACCG";
Sequence orgSeq = new Sequence(Alphabets.DNA, seqData);
BasicDerivedSequence basicDerivedSeq = new BasicDerivedSequence(orgSeq, false, false, 0, 0);
ApplicationLog.WriteLine("Original Sequence: " + basicDerivedSeq.ToString());
BasicDerivedSequence basicDerivedSeqClone = basicDerivedSeq.Clone();
Assert.AreEqual(basicDerivedSeq.ToString(), basicDerivedSeqClone.ToString());
ApplicationLog.WriteLine("Cloned Sequence: " + basicDerivedSeqClone.ToString());
ApplicationLog.WriteLine("BasicDerivedSequenceClone test completed");
}
/// <summary>
/// Writes an ISequence to a FASTA file in the location specified by the writer.
/// </summary>
/// <param name="sequence">The sequence to format.</param>
/// <param name="writer">The TextWriter used to write the formatted sequence text.</param>
public override void Format(ISequence sequence, TextWriter writer)
{
// write header
writer.WriteLine(">" + sequence.ID);
// write sequence
BasicDerivedSequence derivedSeq = new BasicDerivedSequence(sequence, false, false, 0, 0);
for (int lineStart = 0; lineStart < sequence.Count; lineStart += _maxLineLength)
{
derivedSeq.RangeStart = lineStart;
derivedSeq.RangeLength = Math.Min(_maxLineLength, sequence.Count - lineStart);
writer.WriteLine(derivedSeq.ToString());
}
writer.Flush();
}
// writes the sequence to the sepecified writer.
private void WriteSeqData(ISequence sequence, string type, TextWriter writer)
{
if (sequence.Count > 0)
{
type = GetGenericTypeString(sequence.MoleculeType);
WriteHeaderLine(writer, type, sequence.DisplayID);
BasicDerivedSequence derivedSeq = new BasicDerivedSequence(sequence, false, false, 0, 0);
for (int lineStart = 0; lineStart < sequence.Count; lineStart += _maxSequenceSymbolsPerLine)
{
derivedSeq.RangeStart = lineStart;
derivedSeq.RangeLength = Math.Min(_maxSequenceSymbolsPerLine, sequence.Count - lineStart);
WriteHeaderLine(writer, derivedSeq.ToString().ToLower(CultureInfo.InvariantCulture));
}
WriteHeaderLine(writer, "end-" + type);
}
}
//
// Given a list of sequences, create a new list with only the Reverse Complements
// of the original sequences.
static IList <ISequence> ReverseComplementSequenceList(IList <ISequence> sequenceList)
{
List <ISequence> updatedSequenceList = new List <ISequence>();
foreach (ISequence seq in sequenceList)
{
ISequence seqReverseComplement = seq.ReverseComplement;
//
// DISCUSSION:
// Should there be an easily accessed indicator that this is a reversed sequence?
// And should we be able to get the 'base' version even if it is a sub-sequence?
//
BasicDerivedSequence derivedSeq = seqReverseComplement as BasicDerivedSequence;
if (derivedSeq != null)
{
derivedSeq.DisplayID = derivedSeq.DisplayID + " Reverse";
// seqReverseComplement.DisplayID = seqReverseComplement.DisplayID + " Reverse";
}
updatedSequenceList.Add(seqReverseComplement);
}
return(updatedSequenceList);
}
//
// Given a list of sequences, create a new list with the orginal sequence followed
// by the Reverse Complement of that sequence.
static IList <ISequence> AddReverseComplementsToSequenceList(IList <ISequence> sequenceList)
{
List <ISequence> updatedSequenceList = new List <ISequence>();
foreach (ISequence seq in sequenceList)
{
ISequence seqReverseComplement = seq.ReverseComplement;
//
// DISCUSSION:
// Should there be an easily accessed indicator that this is a reversed sequence?
//
BasicDerivedSequence derivedSeq = seqReverseComplement as BasicDerivedSequence;
if (derivedSeq != null)
{
derivedSeq.DisplayID = derivedSeq.DisplayID + " Reverse";
// seqReverseComplement.DisplayID = seqReverseComplement.DisplayID + " Reverse";
}
//seqReverseComplement.ID = seqReverseComplement.ID + " Reverse";
updatedSequenceList.Add(seq);
updatedSequenceList.Add(seqReverseComplement);
}
return(updatedSequenceList);
}
public void TestMsaBenchMarkLargeDataset()
{
// Test on DNA benchmark dataset
ISequenceParser parser = new FastaParser();
string filepath = @"testdata\FASTA\Protein\Balibase\RV913\BOX032.xml.afa";
IList <ISequence> orgSequences = parser.Parse(filepath);
IList <ISequence> sequences = MsaUtils.UnAlign(orgSequences);
int numberOfSequences = orgSequences.Count;
String outputFilePath = @"tempBOX032.xml.afa";
StreamWriter writer = new StreamWriter(outputFilePath, true);
foreach (ISequence sequence in sequences)
{
writer.WriteLine(">" + sequence.ID);
// write sequence
BasicDerivedSequence derivedSeq = new BasicDerivedSequence(sequence, false, false, 0, 0);
for (int lineStart = 0; lineStart < sequence.Count; lineStart += 60)
{
derivedSeq.RangeStart = lineStart;
derivedSeq.RangeLength = Math.Min(60, sequence.Count - lineStart);
writer.WriteLine(derivedSeq.ToString());
}
writer.Flush();
}
writer.Close();
sequences.Clear();
sequences = parser.Parse(outputFilePath);
Console.WriteLine("Original sequences are:");
for (int i = 0; i < numberOfSequences; ++i)
{
Console.WriteLine(sequences[i].ToString());
}
Console.WriteLine("Benchmark sequences are:");
for (int i = 0; i < numberOfSequences; ++i)
{
Console.WriteLine(orgSequences[i].ToString());
}
PAMSAMMultipleSequenceAligner.FasterVersion = false;
PAMSAMMultipleSequenceAligner.UseWeights = false;
PAMSAMMultipleSequenceAligner.UseStageB = true;
PAMSAMMultipleSequenceAligner.NumberOfCores = 2;
int gapOpenPenalty = -13;
int gapExtendPenalty = -5;
int kmerLength = 3;
int numberOfDegrees = 2; //Environment.ProcessorCount;
int numberOfPartitions = 16; // Environment.ProcessorCount * 2;
SimilarityMatrix similarityMatrix = new SimilarityMatrix(SimilarityMatrix.StandardSimilarityMatrix.Blosum62);
DistanceFunctionTypes distanceFunctionName = DistanceFunctionTypes.EuclideanDistance;
UpdateDistanceMethodsTypes hierarchicalClusteringMethodName = UpdateDistanceMethodsTypes.Average;
ProfileAlignerNames profileAlignerName = ProfileAlignerNames.NeedlemanWunschProfileAligner;
ProfileScoreFunctionNames profileProfileFunctionName = ProfileScoreFunctionNames.WeightedInnerProduct;
PAMSAMMultipleSequenceAligner msa = new PAMSAMMultipleSequenceAligner
(sequences, MoleculeType.Protein, kmerLength, distanceFunctionName, hierarchicalClusteringMethodName,
profileAlignerName, profileProfileFunctionName, similarityMatrix, gapOpenPenalty, gapExtendPenalty,
numberOfPartitions, numberOfDegrees);
Console.WriteLine("Benchmark SPS score is: {0}", MsaUtils.MultipleAlignmentScoreFunction(orgSequences, similarityMatrix, gapOpenPenalty, gapExtendPenalty));
Console.WriteLine("Aligned sequences in stage 1: {0}", msa.AlignmentScoreA);
for (int i = 0; i < msa.AlignedSequencesA.Count; ++i)
{
Console.WriteLine(msa.AlignedSequencesA[i].ToString());
}
Console.WriteLine("Alignment score Q is: {0}", MsaUtils.CalculateAlignmentScoreQ(msa.AlignedSequencesA, orgSequences));
Console.WriteLine("Alignment score TC is: {0}", MsaUtils.CalculateAlignmentScoreTC(msa.AlignedSequencesA, orgSequences));
Console.WriteLine("Aligned sequences in stage 2: {0}", msa.AlignmentScoreB);
for (int i = 0; i < msa.AlignedSequencesB.Count; ++i)
{
Console.WriteLine(msa.AlignedSequencesB[i].ToString());
}
Console.WriteLine("Alignment score Q is: {0}", MsaUtils.CalculateAlignmentScoreQ(msa.AlignedSequencesB, orgSequences));
Console.WriteLine("Alignment score TC is: {0}", MsaUtils.CalculateAlignmentScoreTC(msa.AlignedSequencesB, orgSequences));
Console.WriteLine("Aligned sequences in stage 3: {0}", msa.AlignmentScoreC);
for (int i = 0; i < msa.AlignedSequencesC.Count; ++i)
{
Console.WriteLine(msa.AlignedSequencesC[i].ToString());
}
Console.WriteLine("Alignment score Q is: {0}", MsaUtils.CalculateAlignmentScoreQ(msa.AlignedSequencesC, orgSequences));
Console.WriteLine("Alignment score TC is: {0}", MsaUtils.CalculateAlignmentScoreTC(msa.AlignedSequencesC, orgSequences));
Console.WriteLine("Aligned sequences final: {0}", msa.AlignmentScore);
for (int i = 0; i < msa.AlignedSequences.Count; ++i)
{
Console.WriteLine(msa.AlignedSequences[i].ToString());
}
Console.WriteLine("Alignment score Q is: {0}", MsaUtils.CalculateAlignmentScoreQ(msa.AlignedSequences, orgSequences));
Console.WriteLine("Alignment score TC is: {0}", MsaUtils.CalculateAlignmentScoreTC(msa.AlignedSequences, orgSequences));
if (File.Exists(outputFilePath))
{
File.Delete(outputFilePath);
}
}
public void TestListOfByteOnISequence()
{
// Create each sequence type and send to the test method which will call all methods of IList<byte>
Sequence sequence = new Sequence(DnaAlphabet.Instance, "ACT");
sequence.IsReadOnly = false;
TestListOfByteMethods(sequence);
sequence = new Sequence(DnaAlphabet.Instance, "ACT");
sequence.UseEncoding = true;
TestListOfByteMethods(sequence);
BasicDerivedSequence basicDerived = new BasicDerivedSequence(new Sequence(DnaAlphabet.Instance, "ACT"), false, false, 0, sequence.Count);
TestListOfByteMethods(basicDerived);
basicDerived = new BasicDerivedSequence(new Sequence(DnaAlphabet.Instance, "ACT"), false, false, 0, sequence.Count);
basicDerived.UseEncoding = true;
TestListOfByteMethods(basicDerived);
DerivedSequence derived = new DerivedSequence(new Sequence(DnaAlphabet.Instance, "ACT"));
TestListOfByteMethods(derived);
derived = new DerivedSequence(new Sequence(DnaAlphabet.Instance, "ACT"));
derived.UseEncoding = true;
TestListOfByteMethods(derived);
QualitativeSequence qualitative = new QualitativeSequence(DnaAlphabet.Instance, FastQFormatType.Illumina, "ACT");
qualitative.IsReadOnly = false;
TestListOfByteMethods(qualitative);
qualitative = new QualitativeSequence(DnaAlphabet.Instance, FastQFormatType.Illumina, "ACT");
qualitative.UseEncoding = true;
TestListOfByteMethods(qualitative);
SegmentedSequence segmented = new SegmentedSequence(new List <ISequence> {
new Sequence(DnaAlphabet.Instance, "A")
{
IsReadOnly = false
},
new Sequence(DnaAlphabet.Instance, "CT")
{
IsReadOnly = false
}
});
TestListOfByteMethods(segmented);
SparseSequence sparse = new SparseSequence(DnaAlphabet.Instance, 0,
new List <ISequenceItem>
{
DnaAlphabet.Instance.A, DnaAlphabet.Instance.C, DnaAlphabet.Instance.T
});
sparse.IsReadOnly = false;
TestListOfByteMethods(sparse);
// No test code for VirtualSequence
}
/// <summary>
/// Returns a sequence which contains bases from the specified sequence as specified by the location.
/// If the location contains accession then the sequence from the referredSequences which matches the
/// accession of the location will be considered.
///
/// For example,
/// if location is "join(100..200, J00089.1:10..50, J00090.2:30..40)"
/// then bases from 100 to 200 will be considered from the sequence parameter and referredSequences will
/// be searched for the J00089.1 and J00090.2 accession if found then those sequences will be considered
/// for constructing the output sequence.
/// If the referred sequence is not found in the referredSequences then an exception will occur.
/// </summary>
/// <param name="location">Location instance.</param>
/// <param name="sequence">Sequence instance from which the sub sequence has to be returned.</param>
/// <param name="referredSequences">A dictionary containing Accession numbers as keys and Sequences as values, this will be used when
/// the location or sublocations contains accession.</param>
public ISequence GetSubSequence(ILocation location, ISequence sequence, Dictionary <string, ISequence> referredSequences)
{
if (location == null)
{
throw new ArgumentNullException(Resource.ParameterNameLocation);
}
if (sequence == null)
{
throw new ArgumentNullException(Resource.ParameterNameSequence);
}
BasicDerivedSequence basicDerSeq = new BasicDerivedSequence(sequence, false, false, -1, -1);
if (location.Operator == LocationOperator.Complement)
{
if (location.SubLocations.Count > 1)
{
throw new ArgumentException(Resource.ComplementWithMorethanOneSubLocs);
}
if (location.SubLocations.Count > 0)
{
basicDerSeq.Source = location.SubLocations[0].GetSubSequence(sequence, referredSequences);
}
else
{
basicDerSeq.Source = GetSubSequence(location.Start, location.End, location.Accession, location.Separator, sequence, referredSequences);
}
basicDerSeq.Complemented = true;
return(new Sequence(sequence.Alphabet, basicDerSeq.ToString()));
}
if (location.Operator == LocationOperator.Order)
{
List <ISequence> subSequences = new List <ISequence>();
if (location.SubLocations.Count > 0)
{
foreach (ILocation loc in location.SubLocations)
{
subSequences.Add(loc.GetSubSequence(sequence, referredSequences));
}
}
else
{
basicDerSeq.Source = GetSubSequence(location.Start, location.End, location.Accession, location.Separator, sequence, referredSequences);
subSequences.Add(new Sequence(sequence.Alphabet, basicDerSeq.ToString()));
}
return(new SegmentedSequence(subSequences));
}
if (location.Operator == LocationOperator.Join)
{
if (location.SubLocations.Count > 0)
{
List <ISequence> subSequences = new List <ISequence>();
foreach (ILocation loc in location.SubLocations)
{
subSequences.Add(loc.GetSubSequence(sequence, referredSequences));
}
Sequence seq = new Sequence(sequence.Alphabet);
foreach (ISequence subSeq in subSequences)
{
seq.InsertRange(seq.Count, subSeq.ToString());
}
return(seq);
}
else
{
return(GetSubSequence(location.Start, location.End, location.Accession, location.Separator, sequence, referredSequences));
}
}
if (location.SubLocations.Count > 0)
{
throw new ArgumentException(Resource.NoneWithSubLocs);
}
return(GetSubSequence(location.Start, location.End, location.Accession, location.Separator, sequence, referredSequences));
}
// The headers for all sequences go at the top of the file before any features.
private void WriteHeaders(ICollection <ISequence> sequenceList, TextWriter writer)
{
// look for file-scope data tha is common to all sequences; null signifies no match
string source = null;
string version = null;
string type = null;
bool firstSeq = true;
foreach (ISequence sequence in sequenceList)
{
if (firstSeq)
{
// source and version go together; can't output one without the other
if (sequence.Metadata.ContainsKey("source") && sequence.Metadata.ContainsKey("version"))
{
source = sequence.Metadata["source"] as string;
version = sequence.Metadata["version"] as string;
}
// map to generic string; e.g. mRNA, tRNA -> RNA
type = GetGenericTypeString(sequence.MoleculeType);
firstSeq = false;
}
else
{
// source and version go together; can't output one without the other
if (source != null)
{
bool sourceAndVersionMatchOthers =
sequence.Metadata.ContainsKey("source") &&
sequence.Metadata.ContainsKey("version") &&
source == sequence.Metadata["source"] as string &&
version == sequence.Metadata["version"] as string;
// set both to null if this seq source and version don't match previous ones
if (!sourceAndVersionMatchOthers)
{
source = null;
version = null;
}
}
// set type to null if this seq type doesn't match previous types
if (type != null && type != GetGenericTypeString(sequence.MoleculeType))
{
type = null;
}
}
}
// formatting using gff version 2
WriteHeaderLine(writer, "gff-version", "2");
// only output source if they all match
if (source != null)
{
WriteHeaderLine(writer, "source-version", source, version);
}
// today's date
WriteHeaderLine(writer, "date", DateTime.Today.ToString("yyyy-MM-dd"));
// type header
if (type == null)
{
foreach (ISequence sequence in sequenceList)
{
type = GetGenericTypeString(sequence.MoleculeType);
// only ouput seq-specific type header if this seq won't have its type
// output as part of a sequence data header; don't need to output if DNA,
// as DNA is default
if (type != MoleculeType.DNA.ToString() &&
(!ShouldWriteSequenceData || sequence.Count == 0))
{
WriteHeaderLine(writer, "type", type, sequence.DisplayID);
}
}
}
else
{
// output that the types all match; don't need to output if DNA, as DNA is default
if (type != MoleculeType.DNA.ToString())
{
WriteHeaderLine(writer, "type", type);
}
}
// sequence data
if (ShouldWriteSequenceData)
{
foreach (ISequence sequence in sequenceList)
{
if (sequence.Count > 0)
{
type = GetGenericTypeString(sequence.MoleculeType);
WriteHeaderLine(writer, type, sequence.DisplayID);
BasicDerivedSequence derivedSeq = new BasicDerivedSequence(sequence, false, false, 0, 0);
for (int lineStart = 0; lineStart < sequence.Count; lineStart += _maxSequenceSymbolsPerLine)
{
derivedSeq.RangeStart = lineStart;
derivedSeq.RangeLength = Math.Min(_maxSequenceSymbolsPerLine, sequence.Count - lineStart);
WriteHeaderLine(writer, derivedSeq.ToString().ToLower());
}
WriteHeaderLine(writer, "end-" + type);
}
}
}
// sequence-region header
foreach (ISequence sequence in sequenceList)
{
if (sequence.Metadata.ContainsKey("start") && sequence.Metadata.ContainsKey("end"))
{
WriteHeaderLine(writer, "sequence-region", sequence.DisplayID,
sequence.Metadata["start"] as string, sequence.Metadata["end"] as string);
}
}
}