public void TestMUMmerAlignerMultipleMum()
{
string reference = "ATGCGCATCCCCTT";
string search = "GCGCCCCCTA";
Sequence referenceSeq = null;
Sequence searchSeq = null;
referenceSeq = new Sequence(Alphabets.DNA, reference);
searchSeq = new Sequence(Alphabets.DNA, search);
List<ISequence> searchSeqs = new List<ISequence>();
searchSeqs.Add(searchSeq);
MUMmerAligner mummer = new MUMmerAligner();
mummer.LengthOfMUM = 4;
mummer.PairWiseAlgorithm = new NeedlemanWunschAligner();
IList<IPairwiseSequenceAlignment> result = mummer.AlignSimple(referenceSeq, searchSeqs);
// Check if output is not null
Assert.AreNotEqual(null, result);
IList<IPairwiseSequenceAlignment> expectedOutput = new List<IPairwiseSequenceAlignment>();
IPairwiseSequenceAlignment align = new PairwiseSequenceAlignment();
PairwiseAlignedSequence alignedSeq = new PairwiseAlignedSequence();
alignedSeq.FirstSequence = new Sequence(Alphabets.DNA, "ATGCGCATCCCCTT");
alignedSeq.SecondSequence = new Sequence(Alphabets.DNA, "--GCGC--CCCCTA");
alignedSeq.Consensus = new Sequence(AmbiguousDnaAlphabet.Instance, "ATGCGCATCCCCTW");
alignedSeq.Score = -11;
alignedSeq.FirstOffset = 0;
alignedSeq.SecondOffset = 2;
align.PairwiseAlignedSequences.Add(alignedSeq);
expectedOutput.Add(align);
Assert.IsTrue(CompareAlignment(result, expectedOutput));
}
private void ValidateMUMmerAlignGeneralTestCases(string nodeName)
{
// Gets the reference sequence from the configuration file
string filePath = utilityObj.xmlUtil.GetTextValue(nodeName, Constants.FilePathNode);
Assert.IsNotNull(filePath);
ApplicationLog.WriteLine(string.Format(null, "MUMmer P2 : Successfully validated the File Path '{0}'.", filePath));
var fastaParserObj = new FastAParser();
IEnumerable<ISequence> referenceSeqs = fastaParserObj.Parse(filePath);
ISequence referenceSeq = referenceSeqs.ElementAt(0);
// Gets the reference sequence from the configuration file
string queryFilePath = utilityObj.xmlUtil.GetTextValue(nodeName, Constants.SearchSequenceFilePathNode);
Assert.IsNotNull(queryFilePath);
ApplicationLog.WriteLine(string.Format(null, "MUMmer P2 : Successfully validated the Search File Path '{0}'.", queryFilePath));
var fastaParserObj1 = new FastAParser();
IEnumerable<ISequence> querySeqs = fastaParserObj1.Parse(queryFilePath);
string mumLength = utilityObj.xmlUtil.GetTextValue(nodeName, Constants.MUMAlignLengthNode);
var mum = new MUMmerAligner
{
LengthOfMUM = long.Parse(mumLength, null),
StoreMUMs = true,
PairWiseAlgorithm = new NeedlemanWunschAligner(),
GapOpenCost = int.Parse(utilityObj.xmlUtil.GetTextValue(nodeName, Constants.GapOpenCostNode), null)
};
IList<IPairwiseSequenceAlignment> align = mum.Align(referenceSeq, querySeqs);
// Validate FinalMUMs and MUMs Properties.
Assert.IsNotNull(mum.MUMs);
string expectedScore = utilityObj.xmlUtil.GetTextValue(nodeName, Constants.ScoreNodeName);
string[] expectedSequences = utilityObj.xmlUtil.GetTextValues(nodeName, Constants.ExpectedSequencesNode);
IList<IPairwiseSequenceAlignment> expectedOutput = new List<IPairwiseSequenceAlignment>();
IPairwiseSequenceAlignment seqAlign = new PairwiseSequenceAlignment();
var alignedSeq = new PairwiseAlignedSequence
{
FirstSequence = new Sequence(referenceSeq.Alphabet, expectedSequences[0]),
SecondSequence = new Sequence(referenceSeq.Alphabet, expectedSequences[1]),
Score = Convert.ToInt32(expectedScore, null),
FirstOffset = Int32.MinValue,
SecondOffset = Int32.MinValue,
};
seqAlign.PairwiseAlignedSequences.Add(alignedSeq);
expectedOutput.Add(seqAlign);
Assert.IsTrue(AlignmentHelpers.CompareAlignment(align, expectedOutput));
ApplicationLog.WriteLine("MUMmer P2 : Successfully validated the aligned sequences.");
}
/// <summary>
/// Initializes static members of the SequenceAligners class.
/// Static constructor
/// </summary>
static SequenceAligners()
{
NUCmer = new NucmerPairwiseAligner();
MUMmer = new MUMmerAligner();
NeedlemanWunsch = new NeedlemanWunschAligner();
SmithWaterman = new SmithWatermanAligner();
var knownAligners = new List<ISequenceAligner> { SmithWaterman, NeedlemanWunsch, MUMmer, NUCmer };
// Get the registered aligners
IEnumerable<ISequenceAligner> registeredAligners = GetAligners();
if (null != registeredAligners)
{
knownAligners.AddRange(registeredAligners
.Where(aligner => aligner != null
&& knownAligners.All(sa => string.Compare(sa.Name, aligner.Name, StringComparison.OrdinalIgnoreCase) != 0)));
}
All = knownAligners;
}
public void TestMUMmerAlignerSingleMum()
{
const string reference = "TTAATTTTAG";
const string search = "AGTTTAGAG";
ISequence referenceSeq = new Sequence(Alphabets.DNA, reference);
ISequence searchSeq = new Sequence(Alphabets.DNA, search);
var searchSeqs = new List<ISequence> {searchSeq};
MUMmerAligner mummer = new MUMmerAligner
{
LengthOfMUM = 3,
PairWiseAlgorithm = new NeedlemanWunschAligner(),
GapExtensionCost = -2
};
IList<IPairwiseSequenceAlignment> result = mummer.Align(referenceSeq, searchSeqs);
// Check if output is not null
Assert.AreNotEqual(null, result);
IList<IPairwiseSequenceAlignment> expectedOutput = new List<IPairwiseSequenceAlignment>();
IPairwiseSequenceAlignment align = new PairwiseSequenceAlignment();
PairwiseAlignedSequence alignedSeq = new PairwiseAlignedSequence
{
FirstSequence = new Sequence(Alphabets.DNA, "TTAATTTTAG--"),
SecondSequence = new Sequence(Alphabets.DNA, "---AGTTTAGAG"),
Consensus = new Sequence(AmbiguousDnaAlphabet.Instance, "TTAAKTTTAGAG"),
Score = -6,
FirstOffset = 0,
SecondOffset = 3
};
align.PairwiseAlignedSequences.Add(alignedSeq);
expectedOutput.Add(align);
Assert.IsTrue(CompareAlignment(result, expectedOutput));
}
public void TestMUMmerAlignerSingleMumRNA()
{
const string reference = "AUGCUUUUCCCCCCC";
const string search = "UAUAUUUUGG";
MUMmerAligner mummer = new MUMmerAligner
{
LengthOfMUM = 3,
PairWiseAlgorithm = new NeedlemanWunschAligner(),
SimilarityMatrix = new SimilarityMatrix(SimilarityMatrix.StandardSimilarityMatrix.AmbiguousRna),
GapOpenCost = -8,
GapExtensionCost = -2
};
ISequence referenceSeq = new Sequence(Alphabets.RNA, reference);
List<ISequence> searchSeqs = new List<ISequence> { new Sequence(Alphabets.RNA, search) };
IList<IPairwiseSequenceAlignment> result = mummer.Align(referenceSeq, searchSeqs);
// Check if output is not null
Assert.AreNotEqual(0, result.Count);
IList<IPairwiseSequenceAlignment> expectedOutput = new List<IPairwiseSequenceAlignment>();
IPairwiseSequenceAlignment align = new PairwiseSequenceAlignment();
align.PairwiseAlignedSequences.Add(new PairwiseAlignedSequence
{
FirstSequence = new Sequence(Alphabets.RNA, "-AUGCUUUUCCCCCCC"),
SecondSequence = new Sequence(Alphabets.RNA, "UAUA-UUUUGG-----"),
Consensus = new Sequence(AmbiguousRnaAlphabet.Instance, "UAURCUUUUSSCCCCC"),
Score = -14,
FirstOffset = 1,
SecondOffset = 0
});
expectedOutput.Add(align);
AlignmentHelpers.CompareAlignment(result, expectedOutput);
}
public void TestMUMmer3MultipleMumWithCustomMatrix()
{
string reference = "ATGCGCATCCCCTT";
string search = "GCGCCCCCTA";
Sequence referenceSeq = null;
Sequence searchSeq = null;
referenceSeq = new Sequence(Alphabets.DNA, reference);
searchSeq = new Sequence(Alphabets.DNA, search);
List<ISequence> searchSeqs = new List<ISequence>();
searchSeqs.Add(searchSeq);
int[,] customMatrix = new int[256, 256];
customMatrix[(byte)'A', (byte)'A'] = 3;
customMatrix[(byte)'A', (byte)'T'] = -2;
customMatrix[(byte)'A', (byte)'G'] = -2;
customMatrix[(byte)'A', (byte)'c'] = -2;
customMatrix[(byte)'G', (byte)'G'] = 3;
customMatrix[(byte)'G', (byte)'A'] = -2;
customMatrix[(byte)'G', (byte)'T'] = -2;
customMatrix[(byte)'G', (byte)'C'] = -2;
customMatrix[(byte)'T', (byte)'T'] = 3;
customMatrix[(byte)'T', (byte)'A'] = -2;
customMatrix[(byte)'T', (byte)'G'] = -2;
customMatrix[(byte)'T', (byte)'C'] = -2;
customMatrix[(byte)'C', (byte)'C'] = 3;
customMatrix[(byte)'C', (byte)'T'] = -2;
customMatrix[(byte)'C', (byte)'A'] = -2;
customMatrix[(byte)'C', (byte)'G'] = -2;
DiagonalSimilarityMatrix matrix = new DiagonalSimilarityMatrix(3, -2);
int gapOpenCost = -6;
MUMmerAligner mummer = new MUMmerAligner();
mummer.LengthOfMUM = 4;
mummer.PairWiseAlgorithm = new NeedlemanWunschAligner();
mummer.SimilarityMatrix = matrix;
mummer.GapOpenCost = gapOpenCost;
mummer.GapExtensionCost = -2;
IList<IPairwiseSequenceAlignment> result = mummer.AlignSimple(referenceSeq, searchSeqs);
// Check if output is not null
Assert.AreNotEqual(null, result);
IList<IPairwiseSequenceAlignment> expectedOutput = new List<IPairwiseSequenceAlignment>();
IPairwiseSequenceAlignment align = new PairwiseSequenceAlignment();
PairwiseAlignedSequence alignedSeq = new PairwiseAlignedSequence();
alignedSeq.FirstSequence = new Sequence(Alphabets.DNA, "ATGCGCATCCCCTT");
alignedSeq.SecondSequence = new Sequence(Alphabets.DNA, "--GCGC--CCCCTA");
alignedSeq.Consensus = new Sequence(AmbiguousDnaAlphabet.Instance, "ATGCGCATCCCCTW");
alignedSeq.Score = 1;
alignedSeq.FirstOffset = 0;
alignedSeq.SecondOffset = 2;
align.PairwiseAlignedSequences.Add(alignedSeq);
expectedOutput.Add(align);
Assert.IsTrue(CompareAlignment(result, expectedOutput));
}
/// <summary>
/// Validates the Mummer align method for several test cases for the parameters passed.
/// </summary>
/// <param name="nodeName">Node name to be read from xml</param>
/// <param name="isFilePath">Is Sequence saved in File</param>
/// <param name="isAlignList">Is align method to take list?</param>
/// <param name="addParam">Additional parameter</param>
/// Suppress the ParserParam variable CA1801 as this would be reused later.
void ValidateMUMmerAlignGeneralTestCases(string nodeName, bool isFilePath, bool isAlignList, AdditionalParameters addParam)
{
ISequence referenceSeq;
IList<ISequence> querySeqs;
List<ISequence> alignList = null;
if (isFilePath)
{
// Gets the reference sequence from the configuration file
string filePath = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.FilePathNode);
Assert.IsNotNull(filePath);
Assert.IsTrue(File.Exists(filePath));
IEnumerable<ISequence> referenceSeqs;
FastAParser fastaParserObj = new FastAParser();
referenceSeqs = fastaParserObj.Parse(filePath);
referenceSeq = referenceSeqs.FirstOrDefault();
Assert.IsNotNull(referenceSeq);
// Gets the query sequence from the configuration file
string queryFilePath = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.SearchSequenceFilePathNode);
Assert.IsNotNull(queryFilePath);
Assert.IsTrue(File.Exists(queryFilePath));
querySeqs = fastaParserObj.Parse(queryFilePath).ToList();
ISequence querySeq = querySeqs.First();
if (isAlignList)
{
alignList = new List<ISequence> {referenceSeq, querySeq};
}
}
else
{
// Gets the reference sequence from the configuration file
string referenceSequence = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.SequenceNode);
string referenceSeqAlphabet = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.AlphabetNameNode);
referenceSeq = new Sequence(Utility.GetAlphabet(referenceSeqAlphabet), referenceSequence);
string querySequence = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.SearchSequenceNode);
referenceSeqAlphabet = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.SearchSequenceAlphabetNode);
ISequence querySeq = new Sequence(Utility.GetAlphabet(referenceSeqAlphabet), querySequence);
querySeqs = new List<ISequence>();
if (isAlignList)
{
alignList = new List<ISequence> {referenceSeq, querySeq};
}
else
querySeqs.Add(querySeq);
}
// Setup the algorithm
string mumLength = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.MUMAlignLengthNode);
MUMmerAligner mumAlignObj = new MUMmerAligner {LengthOfMUM = long.Parse(mumLength, null), StoreMUMs = true};
switch (addParam)
{
case AdditionalParameters.PerformSimilarityMatrixChange:
mumAlignObj.SimilarityMatrix = new SimilarityMatrix(SimilarityMatrix.StandardSimilarityMatrix.Blosum50);
mumAlignObj.GapOpenCost = int.Parse(this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.GapOpenCostNode), null);
break;
default:
mumAlignObj.GapOpenCost = int.Parse(this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.GapOpenCostNode), null);
break;
}
IEnumerable<ISequence> alignEnumSeqs = alignList;
IList<IPairwiseSequenceAlignment> align = isAlignList
? mumAlignObj.AlignSimple(alignEnumSeqs)
: mumAlignObj.AlignSimple(referenceSeq, querySeqs);
// Validate MUMs Properties
Assert.IsNotNull(mumAlignObj.MUMs);
string expectedScore = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.ScoreNodeName);
string[] expectedSequences = this.utilityObj.xmlUtil.GetTextValues(nodeName, Constants.ExpectedSequencesNode);
IList<IPairwiseSequenceAlignment> expectedOutput = new List<IPairwiseSequenceAlignment>();
// Validate for two aligned sequences and single aligned sequences appropriately
if (querySeqs.Count <= 1)
{
IPairwiseSequenceAlignment seqAlign = new PairwiseSequenceAlignment();
PairwiseAlignedSequence alignedSeq = new PairwiseAlignedSequence
{
FirstSequence = new Sequence(referenceSeq.Alphabet, expectedSequences[0]),
SecondSequence = new Sequence(referenceSeq.Alphabet, expectedSequences[1]),
Score = Convert.ToInt32(expectedScore,null),
FirstOffset = Int32.MinValue,
SecondOffset = Int32.MinValue,
};
seqAlign.PairwiseAlignedSequences.Add(alignedSeq);
expectedOutput.Add(seqAlign);
Assert.IsTrue(CompareAlignment(align, expectedOutput));
}
else
{
string[] expectedScores = expectedScore.Split(',');
IPairwiseSequenceAlignment seq1Align = new PairwiseSequenceAlignment();
IPairwiseSequenceAlignment seq2Align = new PairwiseSequenceAlignment();
// Get the first sequence for validation
PairwiseAlignedSequence alignedSeq1 = new PairwiseAlignedSequence
{
FirstSequence = new Sequence(referenceSeq.Alphabet, expectedSequences[0]),
SecondSequence = new Sequence(referenceSeq.Alphabet, expectedSequences[1]),
Score = int.Parse(expectedScores[0], null),
FirstOffset = Int32.MinValue,
SecondOffset = Int32.MinValue,
};
seq1Align.PairwiseAlignedSequences.Add(alignedSeq1);
expectedOutput.Add(seq1Align);
// Get the second sequence for validation
PairwiseAlignedSequence alignedSeq2 = new PairwiseAlignedSequence
{
FirstSequence = new Sequence(referenceSeq.Alphabet, expectedSequences[2]),
SecondSequence = new Sequence(referenceSeq.Alphabet, expectedSequences[3]),
Score = int.Parse(expectedScores[1], null),
FirstOffset = Int32.MinValue,
SecondOffset = Int32.MinValue,
};
seq2Align.PairwiseAlignedSequences.Add(alignedSeq2);
expectedOutput.Add(seq2Align);
Assert.IsTrue(CompareAlignment(align, expectedOutput));
}
}
