/// <summary>
/// Constructor for SmithWatermanProfileAligner Aligner.
/// Sets default similarity matrix, gap penalties, and profile function name.
/// Users will typically reset these using parameters specific to their particular sequences and needs.
/// </summary>
/// <param name="similarityMatrix">similarity matrix</param>
/// <param name="profileScoreFunctionName">enum: profileScoreFunctionName</param>
/// <param name="gapOpenPenalty">negative integer</param>
/// <param name="gapExtensionPenalty">negative integer</param>
/// <param name="numberOfPartitions">positive integer</param>
public SmithWatermanProfileAlignerParallel(SimilarityMatrix similarityMatrix,
ProfileScoreFunctionNames profileScoreFunctionName,
int gapOpenPenalty,
int gapExtensionPenalty,
int numberOfPartitions)
: base(similarityMatrix, profileScoreFunctionName, gapOpenPenalty, gapExtensionPenalty, numberOfPartitions)
{
}
/// <summary>
/// Constructor for NeedlemanWunschProfile Aligner.
/// Sets default similarity matrix, gap penalties, and profile function name.
/// Users will typically reset these using parameters specific to their particular sequences and needs.
/// </summary>
/// <param name="similarityMatrix">similarity matrix</param>
/// <param name="profileScoreFunctionName">enum: profileScoreFunctionName</param>
/// <param name="gapOpenPenalty">negative integer</param>
/// <param name="gapExtensionPenalty">negative integer</param>
/// <param name="numberOfPartitions">positive integer</param>
public NeedlemanWunschProfileAlignerSerial(SimilarityMatrix similarityMatrix,
ProfileScoreFunctionNames profileScoreFunctionName,
int gapOpenPenalty,
int gapExtensionPenalty,
int numberOfPartitions)
: base(similarityMatrix, profileScoreFunctionName, gapOpenPenalty, gapExtensionPenalty, numberOfPartitions)
{
}
public static float PairWiseScoreFunction(ISequence sequenceA, ISequence sequenceB, SimilarityMatrix similarityMatrix,
int gapOpenPenalty, int gapExtensionPenalty)
{
if (sequenceA.Count != sequenceB.Count)
{
throw new Exception("Unaligned sequences");
}
float result = 0;
bool isGapA = false;
bool isGapB = false;
for (int i = 0; i < sequenceA.Count; ++i)
{
if (sequenceA.Alphabet.CheckIsGap(sequenceA[i]) && sequenceB.Alphabet.CheckIsGap(sequenceB[i]))
{
continue;
}
if (sequenceA.Alphabet.CheckIsGap(sequenceA[i]) && !sequenceB.Alphabet.CheckIsGap(sequenceB[i]))
{
if (isGapB)
{
isGapB = false;
}
if (isGapA)
{
result += gapExtensionPenalty;
}
else
{
result += gapOpenPenalty;
isGapA = true;
}
continue;
}
if (!sequenceA.Alphabet.CheckIsGap(sequenceA[i]) && sequenceB.Alphabet.CheckIsGap(sequenceB[i]))
{
if (isGapA)
{
isGapA = false;
}
if (isGapB)
{
result += gapExtensionPenalty;
}
else
{
result += gapOpenPenalty;
isGapB = true;
}
continue;
}
result += similarityMatrix[sequenceA[i], sequenceB[i]];
}
return result;
}
/// <summary>
/// Initializes a new instance of the DynamicProgrammingPairwiseAligner class.
/// Constructor for all the pairwise aligner (NeedlemanWunsch, SmithWaterman, Overlap).
/// Sets default similarity matrix and gap penalties.
/// Users will typically reset these using parameters specific to their particular sequences and needs.
/// </summary>
protected DynamicProgrammingPairwiseAligner()
{
// Set default similarity matrix and gap penalty.
// User will typically choose their own parameters, these defaults are reasonable for many cases.
// Molecule type is set to protein, since this will also work for DNA and RNA in the
// special case of a diagonal similarity matrix.
this.InternalSimilarityMatrix = new DiagonalSimilarityMatrix(2, -2);
GapOpenCost = -8;
GapExtensionCost = -1;
}
public void PairwiseOverlapProteinSeqAffineGap()
{
string sequenceString1 = "HEAGAWGHEE";
string sequenceString2 = "PAWHEAE";
Sequence sequence1 = new Sequence(Alphabets.Protein, sequenceString1);
Sequence sequence2 = new Sequence(Alphabets.Protein, sequenceString2);
SimilarityMatrix sm = new SimilarityMatrix(SimilarityMatrix.StandardSimilarityMatrix.Blosum50);
int gapPenalty = -8;
PairwiseOverlapAligner overlap = new PairwiseOverlapAligner();
overlap.SimilarityMatrix = sm;
overlap.GapOpenCost = gapPenalty;
overlap.GapExtensionCost = -1;
IList<IPairwiseSequenceAlignment> result = overlap.Align(sequence1, sequence2);
ApplicationLog.WriteLine(string.Format((IFormatProvider)null,
"{0}, Affine; Matrix {1}; GapOpenCost {2}; GapExtenstionCost {3}",
overlap.Name, overlap.SimilarityMatrix.Name, overlap.GapOpenCost, overlap.GapExtensionCost));
foreach (IPairwiseSequenceAlignment sequenceResult in result)
{
ApplicationLog.WriteLine(string.Format((IFormatProvider)null,
"score {0}", sequenceResult.PairwiseAlignedSequences[0].Score));
ApplicationLog.WriteLine(string.Format((IFormatProvider)null,
"input 0 {0}", sequenceResult.FirstSequence.ToString()));
ApplicationLog.WriteLine(string.Format((IFormatProvider)null,
"input 1 {0}", sequenceResult.SecondSequence.ToString()));
ApplicationLog.WriteLine(string.Format((IFormatProvider)null,
"result 0 {0}", sequenceResult.PairwiseAlignedSequences[0].FirstSequence.ToString()));
ApplicationLog.WriteLine(string.Format((IFormatProvider)null,
"result 1 {0}", sequenceResult.PairwiseAlignedSequences[0].SecondSequence.ToString()));
ApplicationLog.WriteLine(string.Format((IFormatProvider)null,
"consesus {0}", sequenceResult.PairwiseAlignedSequences[0].Consensus.ToString()));
}
IList<IPairwiseSequenceAlignment> expectedOutput = new List<IPairwiseSequenceAlignment>();
IPairwiseSequenceAlignment align = new PairwiseSequenceAlignment();
PairwiseAlignedSequence alignedSeq = new PairwiseAlignedSequence();
alignedSeq.FirstSequence = new Sequence(Alphabets.Protein, "GAWGHEE");
alignedSeq.SecondSequence = new Sequence(Alphabets.Protein, "PAW-HEA");
alignedSeq.Consensus = new Sequence(Alphabets.AmbiguousProtein, "XAWGHEX");
alignedSeq.Score = 25;
alignedSeq.FirstOffset = 0;
alignedSeq.SecondOffset = 3;
align.PairwiseAlignedSequences.Add(alignedSeq);
expectedOutput.Add(align);
Assert.IsTrue(CompareAlignment(result, expectedOutput));
}
/// <summary>
///
/// </summary>
public NucmerPairwiseAligner()
{
// Set the default Similarity Matrix
SimilarityMatrix = new SimilarityMatrix(
SimilarityMatrix.StandardSimilarityMatrix.DiagonalScoreMatrix);
// Set the defaults
GapOpenCost = DefaultGapOpenCost;
GapExtensionCost = DefaultGapExtensionCost;
LengthOfMUM = DefaultLengthOfMUM;
// Set the ClusterBuilder properties to defaults
FixedSeparation = ClusterBuilder.DefaultFixedSeparation;
MaximumSeparation = ClusterBuilder.DefaultMaximumSeparation;
MinimumScore = ClusterBuilder.DefaultMinimumScore;
SeparationFactor = ClusterBuilder.DefaultSeparationFactor;
BreakLength = ModifiedSmithWaterman.DefaultBreakLength;
}
/// <summary>
/// Aligns multiple sequences using a multiple sequence aligner.
/// This sample uses PAMSAM with a set of default parameters.
/// </summary>
/// <param name="sequences">List of sequences to align.</param>
/// <returns>List of ISequenceAlignment</returns>
public static IList<ISequence> DoMultipleSequenceAlignment(List<ISequence> sequences)
{
// Initialize objects for constructor
SimilarityMatrix similarityMatrix = new SimilarityMatrix(SimilarityMatrix.StandardSimilarityMatrix.AmbiguousDna);
int gapOpenPenalty = -4;
int gapExtendPenalty = -1;
int kmerLength = 3;
DistanceFunctionTypes distanceFunctionName = DistanceFunctionTypes.EuclideanDistance;
UpdateDistanceMethodsTypes hierarchicalClusteringMethodName = UpdateDistanceMethodsTypes.Average;
ProfileAlignerNames profileAlignerName = ProfileAlignerNames.NeedlemanWunschProfileAligner;
ProfileScoreFunctionNames profileProfileFunctionName = ProfileScoreFunctionNames.WeightedInnerProduct;
// Call aligner
PAMSAMMultipleSequenceAligner msa = new PAMSAMMultipleSequenceAligner
(sequences, kmerLength, distanceFunctionName, hierarchicalClusteringMethodName,
profileAlignerName, profileProfileFunctionName, similarityMatrix, gapOpenPenalty, gapExtendPenalty,
Environment.ProcessorCount * 2, Environment.ProcessorCount);
return msa.AlignedSequences;
}
/// <summary>
/// Construct a progressive aligner
/// </summary>
/// <param name="profileAlignerName">ProfileAlignerNames member</param>
/// <param name="similarityMatrix">similarity matrix</param>
/// <param name="gapOpenPenalty">negative gapOpenPenalty</param>
/// <param name="gapExtendPenalty">negative gapExtendPenalty</param>
public ProgressiveAligner(ProfileAlignerNames profileAlignerName,
SimilarityMatrix similarityMatrix,
int gapOpenPenalty,
int gapExtendPenalty)
{
// Get ProfileAligner ready
switch (profileAlignerName)
{
case (ProfileAlignerNames.NeedlemanWunschProfileAligner):
_profileAligner = new NeedlemanWunschProfileAlignerSerial();
break;
case (ProfileAlignerNames.SmithWatermanProfileAligner):
_profileAligner = new SmithWatermanProfileAlignerSerial();
break;
default:
throw new Exception("Invalid profile aligner name");
}
_profileAligner.SimilarityMatrix = similarityMatrix;
_profileAligner.GapOpenCost = gapOpenPenalty;
_profileAligner.GapExtensionCost = gapExtendPenalty;
_alignedSequences = new List<ISequence>();
}
/// <summary>
/// Symmetrized entropy of observation vectors
/// </summary>
/// <param name="similarityMatrix">similarity matrix</param>
/// <param name="profileIndexA">the first profile vector (normalized)</param>
/// <param name="profileIndexB">the second profile vector (normalized)</param>
protected float SymmetrizedEntropy(SimilarityMatrix similarityMatrix, int profileIndexA, int profileIndexB)
{
return MsaUtils.SymmetrizedEntropy(_profileAlignmentA.ProfilesMatrix[profileIndexA], _profileAlignmentB.ProfilesMatrix[profileIndexB]);
}
/// <summary>
/// Validate the inputs.
/// </summary>
/// <param name="referenceSequence">Reference sequence.</param>
/// <param name="querySequenceList">List of input sequences.</param>
/// <returns>Are inputs valid.</returns>
private bool Validate(
ISequence referenceSequence,
IEnumerable<ISequence> querySequenceList)
{
bool isValidLength = false;
if (null == referenceSequence)
{
string message = Properties.Resource.ReferenceSequenceCannotBeNull;
Debug.WriteLine(message);
throw new ArgumentNullException("referenceSequence");
}
if (null == querySequenceList)
{
string message = Properties.Resource.QueryListCannotBeNull;
Debug.WriteLine(message);
throw new ArgumentNullException("querySequenceList");
}
if ((referenceSequence.Alphabet != Alphabets.DNA) && (referenceSequence.Alphabet != Alphabets.RNA))
{
string message = string.Format(
CultureInfo.CurrentCulture,
Properties.Resource.OnlyDNAOrRNAInput,
"MUMmer");
Debug.WriteLine(message);
throw new ArgumentException(message, "referenceSequence");
}
// setting default similarity matrix based on DNA or RNA
if (SimilarityMatrix == null)
{
if (referenceSequence.Alphabet == Alphabets.RNA)
{
SimilarityMatrix = new SimilarityMatrix(SimilarityMatrix.StandardSimilarityMatrix.AmbiguousRna);
}
else if (referenceSequence.Alphabet == Alphabets.DNA)
{
SimilarityMatrix = new SimilarityMatrix(SimilarityMatrix.StandardSimilarityMatrix.AmbiguousDna);
}
}
if (!SimilarityMatrix.ValidateSequence(referenceSequence))
{
string message = Properties.Resource.FirstInputSequenceMismatchSimilarityMatrix;
Debug.WriteLine(message);
throw new ArgumentException(message, "referenceSequence");
}
if (referenceSequence.Count < this.LengthOfMUM)
{
string message = String.Format(
CultureInfo.CurrentCulture,
Properties.Resource.InputSequenceMustBeGreaterThanMUM,
this.LengthOfMUM);
Debug.WriteLine(message);
throw new ArgumentException(message, "referenceSequence");
}
foreach (ISequence querySequence in querySequenceList)
{
if (null == querySequence)
{
string message = Properties.Resource.QuerySequenceCannotBeNull;
Debug.WriteLine(message);
throw new ArgumentNullException("querySequenceList", message);
}
if (referenceSequence.Alphabet != querySequence.Alphabet)
{
string message = Properties.Resource.InputAlphabetsMismatch;
Debug.WriteLine(message);
throw new ArgumentException(message);
}
if (!SimilarityMatrix.ValidateSequence(querySequence))
{
string message = Properties.Resource.SecondInputSequenceMismatchSimilarityMatrix;
Debug.WriteLine(message);
throw new ArgumentException(message, "querySequenceList");
}
if (querySequence.Count >= this.LengthOfMUM)
{
isValidLength = true;
}
}
if (!isValidLength)
{
string message = String.Format(
CultureInfo.CurrentCulture,
Properties.Resource.InputSequenceMustBeGreaterThanMUM,
this.LengthOfMUM);
Debug.WriteLine(message);
throw new ArgumentException(message, "querySequenceList");
}
if (1 > this.LengthOfMUM)
{
string message = Properties.Resource.MUMLengthTooSmall;
Debug.WriteLine(message);
throw new ArgumentException(message);
}
return true;
}
/// <summary>
/// Validates PairwiseOverlapAlignment algorithm for the parameters passed.
/// </summary>
/// <param name="nodeName">Xml node name</param>
/// <param name="isTextFile">Is text file an input.</param>
/// <param name="caseType">Case Type</param>
/// <param name="additionalParameter">parameter based on which certain validations are done.</param>
/// <param name="alignType">Is the Align type Simple or Align with Gap Extension cost?</param>
/// <param name="similarityMatrixParam">Similarity Matrix</param>
private void ValidatePairwiseOverlapAlignment(string nodeName, bool isTextFile,
SequenceCaseType caseType, AlignParameters additionalParameter,
AlignmentType alignType,
SimilarityMatrixParameters similarityMatrixParam)
{
Sequence aInput = null;
Sequence bInput = null;
IAlphabet alphabet = Utility.GetAlphabet(this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.AlphabetNameNode));
if (isTextFile)
{
// Read the xml file for getting both the files for aligning.
string filePath1 = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.FilePathNode1);
string filePath2 = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.FilePathNode2);
var parser1 = new FastAParser();
ISequence originalSequence1 = parser1.Parse(filePath1).ElementAt(0);
ISequence originalSequence2 = parser1.Parse(filePath2).ElementAt(0);
// Create input sequence for sequence string in different cases.
GetSequenceWithCaseType(new string(originalSequence1.Select(a => (char) a).ToArray()),
new string(originalSequence2.Select(a => (char) a).ToArray()), alphabet,
caseType, out aInput, out bInput);
}
else
{
string originalSequence1 = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.SequenceNode1);
string originalSequence2 = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.SequenceNode2);
// Create input sequence for sequence string in different cases.
GetSequenceWithCaseType(
originalSequence1,
originalSequence2,
alphabet,
caseType,
out aInput,
out bInput);
}
var aInputString = new string(aInput.Select(a => (char) a).ToArray());
var bInputString = new string(bInput.Select(a => (char) a).ToArray());
ApplicationLog.WriteLine(string.Format(null,
"PairwiseOverlapAligner P2 : First sequence used is '{0}'.",
aInputString));
ApplicationLog.WriteLine(string.Format(null,
"PairwiseOverlapAligner P2 : Second sequence used is '{0}'.",
bInputString));
// Create similarity matrix object for a given file.
string blosumFilePath = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.BlosumFilePathNode);
SimilarityMatrix sm = null;
switch (similarityMatrixParam)
{
case SimilarityMatrixParameters.TextReader:
using (TextReader reader = new StreamReader(blosumFilePath))
sm = new SimilarityMatrix(reader);
break;
case SimilarityMatrixParameters.DiagonalMatrix:
string matchValue = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.MatchScoreNode);
string misMatchValue = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.MisMatchScoreNode);
sm = new DiagonalSimilarityMatrix(int.Parse(matchValue, null),
int.Parse(misMatchValue, null));
break;
default:
sm = new SimilarityMatrix(new StreamReader(blosumFilePath));
break;
}
int gapOpenCost = int.Parse(this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.GapOpenCostNode), null);
int gapExtensionCost = int.Parse(this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.GapExtensionCostNode), null);
// Create PairwiseOverlapAligner instance and set its values.
var pairwiseOverlapObj = new PairwiseOverlapAligner();
if (additionalParameter != AlignParameters.AllParam)
{
pairwiseOverlapObj.SimilarityMatrix = sm;
pairwiseOverlapObj.GapOpenCost = gapOpenCost;
pairwiseOverlapObj.GapExtensionCost = gapExtensionCost;
}
IList<IPairwiseSequenceAlignment> result = null;
// Align the input sequences.
switch (additionalParameter)
{
case AlignParameters.AlignList:
var sequences = new List<ISequence>();
sequences.Add(aInput);
sequences.Add(bInput);
switch (alignType)
{
case AlignmentType.Align:
result = pairwiseOverlapObj.Align(sequences);
break;
default:
result = pairwiseOverlapObj.AlignSimple(sequences);
break;
}
break;
case AlignParameters.AlignTwo:
switch (alignType)
{
case AlignmentType.Align:
result = pairwiseOverlapObj.Align(aInput, bInput);
break;
default:
result = pairwiseOverlapObj.AlignSimple(aInput, bInput);
break;
}
break;
case AlignParameters.AllParam:
switch (alignType)
{
case AlignmentType.Align:
result = pairwiseOverlapObj.Align(sm, gapOpenCost,
gapExtensionCost, aInput, bInput);
break;
default:
result = pairwiseOverlapObj.AlignSimple(sm, gapOpenCost, aInput, bInput);
break;
}
break;
default:
break;
}
aInput = null;
bInput = null;
sm = null;
// Get the expected sequence and scorde from xml config.
string expectedSequence1 = string.Empty;
string expectedSequence2 = string.Empty;
string expectedScore = string.Empty;
switch (alignType)
{
case AlignmentType.Align:
expectedScore = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.ExpectedGapExtensionScoreNode);
expectedSequence1 = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.ExpectedGapExtensionSequence1Node);
expectedSequence2 = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.ExpectedGapExtensionSequence2Node);
break;
default:
expectedScore = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.ExpectedScoreNode);
expectedSequence1 = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.ExpectedSequenceNode1);
expectedSequence2 = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.ExpectedSequenceNode2);
break;
}
IList<IPairwiseSequenceAlignment> expectedOutput = new List<IPairwiseSequenceAlignment>();
string[] expectedSequences1, expectedSequences2;
var seperators = new char[1] {';'};
expectedSequences1 = expectedSequence1.Split(seperators);
expectedSequences2 = expectedSequence2.Split(seperators);
IPairwiseSequenceAlignment align = new PairwiseSequenceAlignment();
PairwiseAlignedSequence alignedSeq;
for (int i = 0; i < expectedSequences1.Length; i++)
{
alignedSeq = new PairwiseAlignedSequence
{
FirstSequence = new Sequence(alphabet, expectedSequences1[i]),
SecondSequence = new Sequence(alphabet, expectedSequences2[i]),
Score = Convert.ToInt32(expectedScore, null),
FirstOffset = Int32.MinValue,
SecondOffset = Int32.MinValue,
};
align.PairwiseAlignedSequences.Add(alignedSeq);
}
expectedOutput.Add(align);
Assert.IsTrue(AlignmentHelpers.CompareAlignment(result, expectedOutput,true));
ApplicationLog.WriteLine(string.Format(null, "PairwiseOverlapAligner P2 : Final Score '{0}'.", expectedScore));
ApplicationLog.WriteLine(string.Format(null, "PairwiseOverlapAligner P2 : Aligned First Sequence is '{0}'.", expectedSequence1));
ApplicationLog.WriteLine(string.Format(null, "PairwiseOverlapAligner P2 : Aligned Second Sequence is '{0}'.", expectedSequence2));
}
/// <summary>
/// Pairwise alignment of two sequences using an affine gap penalty. The various algorithms in derived classes (NeedlemanWunsch,
/// SmithWaterman, and PairwiseOverlap) all use this general engine for alignment with an affine gap penalty.
/// </summary>
/// <param name="localSimilarityMatrix">Scoring matrix.</param>
/// <param name="gapOpenPenalty">Gap open penalty (by convention, use a negative number for this.).</param>
/// <param name="gapExtensionPenalty">Gap extension penalty (by convention, use a negative number for this.).</param>
/// <param name="inputA">First input sequence.</param>
/// <param name="inputB">Second input sequence.</param>
/// <returns>A list of sequence alignments.</returns>
public IList<IPairwiseSequenceAlignment> Align(
SimilarityMatrix localSimilarityMatrix,
int gapOpenPenalty,
int gapExtensionPenalty,
ISequence inputA,
ISequence inputB)
{
this.SimilarityMatrix = localSimilarityMatrix;
this.GapOpenCost = gapOpenPenalty;
this.GapExtensionCost = gapExtensionPenalty;
return DoAlign(inputA, inputB, true);
}
/// <summary>
/// Inializes a new alignment job
/// </summary>
/// <param name="similarityMatrix"></param>
/// <param name="gapOpenCost"></param>
/// <param name="aInput"></param>
/// <param name="bInput"></param>
public PairwiseOverlapSimpleAlignmentJob(SimilarityMatrix similarityMatrix, int gapOpenCost, ISequence aInput, ISequence bInput)
: base(similarityMatrix, gapOpenCost, 0, aInput, bInput) { }
private void InValidateSmithWatermanAlignmentWithInvalidSequence(string nodeName,
bool isTextFile,
InvalidSequenceType invalidSequenceType,
AlignParameters additionalParameter,
AlignmentType alignType,
InvalidSequenceType sequenceType)
{
IAlphabet alphabet = Utility.GetAlphabet(this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.AlphabetNameNode));
Exception actualException = null;
Sequence aInput = null;
Sequence bInput = null;
if (isTextFile)
{
// Read the xml file for getting both the files for aligning.
string filepath = this.GetInputFileNameWithInvalidType(nodeName, invalidSequenceType);
// Create input sequence for sequence string in different cases.
try
{
// Parse the files and get the sequence.
IEnumerable<ISequence> seqs = null;
var parser = new FastAParser();
seqs = parser.Parse(filepath);
aInput = new Sequence(alphabet, new string(seqs.ElementAt(0).Select(a => (char) a).ToArray()));
}
catch (Exception ex)
{
actualException = ex;
}
}
else
{
string originalSequence = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.InvalidSequence1);
// Create input sequence for sequence string in different cases.
try
{
aInput = new Sequence(alphabet, originalSequence);
}
catch (ArgumentException ex)
{
actualException = ex;
}
}
if (actualException == null)
{
bInput = aInput;
// Create similarity matrix object for a given file.
string blosumFilePath = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.BlosumFilePathNode);
var sm = new SimilarityMatrix(new StreamReader(blosumFilePath));
int gapOpenCost = int.Parse(this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.GapOpenCostNode), null);
int gapExtensionCost = int.Parse(this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.GapExtensionCostNode), null);
// Create SmithWatermanAligner instance and set its values.
var smithWatermanObj = new SmithWatermanAligner();
if (additionalParameter != AlignParameters.AllParam)
{
smithWatermanObj.SimilarityMatrix = sm;
smithWatermanObj.GapOpenCost = gapOpenCost;
smithWatermanObj.GapExtensionCost = gapExtensionCost;
}
// Align the input sequences and catch the exception.
switch (additionalParameter)
{
case AlignParameters.AlignList:
switch (alignType)
{
case AlignmentType.Align:
try
{
smithWatermanObj.Align(new List<ISequence> {aInput, bInput});
}
catch (ArgumentException ex)
{
actualException = ex;
}
break;
default:
try
{
smithWatermanObj.AlignSimple(new List<ISequence> {aInput, bInput});
}
catch (ArgumentException ex)
{
actualException = ex;
}
break;
}
break;
case AlignParameters.AlignTwo:
switch (alignType)
{
case AlignmentType.Align:
try
{
smithWatermanObj.Align(aInput, bInput);
}
catch (ArgumentException ex)
{
actualException = ex;
}
break;
default:
try
{
smithWatermanObj.AlignSimple(aInput, bInput);
}
catch (ArgumentException ex)
{
actualException = ex;
}
break;
}
break;
case AlignParameters.AllParam:
switch (alignType)
{
case AlignmentType.Align:
try
{
smithWatermanObj.Align(sm, gapOpenCost,
gapExtensionCost, aInput, bInput);
}
catch (ArgumentException ex)
{
actualException = ex;
}
break;
default:
try
{
smithWatermanObj.AlignSimple(sm, gapOpenCost,
aInput, bInput);
}
catch (ArgumentException ex)
{
actualException = ex;
}
break;
}
break;
default:
break;
}
}
// Validate Error messages for Invalid Sequence types.
string expectedErrorMessage = this.GetExpectedErrorMeesageWithInvalidSequenceType(nodeName,
sequenceType);
Assert.AreEqual(expectedErrorMessage, actualException.Message);
ApplicationLog.WriteLine(string.Concat(
"SmithWatermanAligner P2 : Expected Error message is thrown ", expectedErrorMessage));
}
/// <summary>
/// Validates PairwiseOverlapAlignment algorithm for the parameters passed.
/// </summary>
/// <param name="nodeName">Node Name in the xml.</param>
/// <param name="alignParam">parameter based on which certain validations are done.</param>
/// <param name="similarityMatrixParam">Similarity Matrix Parameter.</param>
/// <param name="alignType">Alignment Type</param>
private void ValidatePairwiseOverlapAlignment(string nodeName, AlignParameters alignParam,
SimilarityMatrixParameters similarityMatrixParam,
AlignmentType alignType)
{
ISequence aInput;
ISequence bInput;
IAlphabet alphabet = Utility.GetAlphabet(this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.AlphabetNameNode));
if (alignParam.ToString().Contains("Code"))
{
string sequence1 = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.SequenceNode1);
string sequence2 = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.SequenceNode2);
aInput = new Sequence(alphabet, sequence1);
bInput = new Sequence(alphabet, sequence2);
}
else
{
// Read the xml file for getting both the files for aligning.
string filePath1 = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.FilePathNode1);
string filePath2 = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.FilePathNode2);
var parser1 = new FastAParser { Alphabet = alphabet };
aInput = parser1.Parse(filePath1).ElementAt(0);
bInput = parser1.Parse(filePath2).ElementAt(0);
}
string blosumFilePath = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.BlosumFilePathNode);
SimilarityMatrix sm;
switch (similarityMatrixParam)
{
case SimilarityMatrixParameters.TextReader:
using (TextReader reader = new StreamReader(blosumFilePath))
sm = new SimilarityMatrix(reader);
break;
case SimilarityMatrixParameters.DiagonalMatrix:
string matchValue = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.MatchScoreNode);
string misMatchValue = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.MisMatchScoreNode);
sm = new DiagonalSimilarityMatrix(int.Parse(matchValue, null), int.Parse(misMatchValue, null));
break;
default:
sm = new SimilarityMatrix(new StreamReader(blosumFilePath));
break;
}
int gapOpenCost = int.Parse(this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.GapOpenCostNode), null);
int gapExtensionCost = int.Parse(this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.GapExtensionCostNode), null);
var pairwiseOverlapObj = new PairwiseOverlapAligner();
if (AlignParameters.AllParam != alignParam)
{
pairwiseOverlapObj.SimilarityMatrix = sm;
pairwiseOverlapObj.GapOpenCost = gapOpenCost;
}
IList<IPairwiseSequenceAlignment> result = null;
switch (alignParam)
{
case AlignParameters.AlignList:
case AlignParameters.AlignListCode:
var sequences = new List<ISequence> {aInput, bInput};
switch (alignType)
{
case AlignmentType.Align:
pairwiseOverlapObj.GapExtensionCost = gapExtensionCost;
result = pairwiseOverlapObj.Align(sequences);
break;
default:
result = pairwiseOverlapObj.AlignSimple(sequences);
break;
}
break;
case AlignParameters.AllParam:
case AlignParameters.AllParamCode:
switch (alignType)
{
case AlignmentType.Align:
pairwiseOverlapObj.GapExtensionCost = gapExtensionCost;
result = pairwiseOverlapObj.Align(sm, gapOpenCost, gapExtensionCost, aInput, bInput);
break;
default:
result = pairwiseOverlapObj.AlignSimple(sm, gapOpenCost, aInput, bInput);
break;
}
break;
case AlignParameters.AlignTwo:
case AlignParameters.AlignTwoCode:
switch (alignType)
{
case AlignmentType.Align:
pairwiseOverlapObj.GapExtensionCost = gapExtensionCost;
result = pairwiseOverlapObj.Align(aInput, bInput);
break;
default:
result = pairwiseOverlapObj.AlignSimple(aInput, bInput);
break;
}
break;
default:
break;
}
// Read the xml file for getting both the files for aligning.
string expectedSequence1;
string expectedSequence2;
string expectedScore;
switch (alignType)
{
case AlignmentType.Align:
expectedScore = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.ExpectedGapExtensionScoreNode);
expectedSequence1 = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.ExpectedGapExtensionSequence1Node);
expectedSequence2 = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.ExpectedGapExtensionSequence2Node);
break;
default:
expectedScore = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.ExpectedScoreNode);
expectedSequence1 = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.ExpectedSequenceNode1);
expectedSequence2 = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.ExpectedSequenceNode2);
break;
}
IList<IPairwiseSequenceAlignment> expectedOutput = new List<IPairwiseSequenceAlignment>();
var seperators = new [] {';'};
string[] expectedSequences1 = expectedSequence1.Split(seperators);
string[] expectedSequences2 = expectedSequence2.Split(seperators);
IPairwiseSequenceAlignment align = new PairwiseSequenceAlignment();
for (int i = 0; i < expectedSequences1.Length; i++)
{
PairwiseAlignedSequence alignedSeq = new PairwiseAlignedSequence
{
FirstSequence = new Sequence(alphabet, expectedSequences1[i]),
SecondSequence = new Sequence(alphabet, expectedSequences2[i]),
Score = Convert.ToInt32(expectedScore, null),
FirstOffset = Int32.MinValue,
SecondOffset = Int32.MinValue,
};
align.PairwiseAlignedSequences.Add(alignedSeq);
}
expectedOutput.Add(align);
Assert.IsTrue(AlignmentHelpers.CompareAlignment(result, expectedOutput, true));
ApplicationLog.WriteLine(string.Format(null, "PairwiseOverlapAligner P1 : Final Score '{0}'.", expectedScore));
ApplicationLog.WriteLine(string.Format(null, "PairwiseOverlapAligner P1 : Aligned First Sequence is '{0}'.", expectedSequence1));
ApplicationLog.WriteLine(string.Format(null, "PairwiseOverlapAligner P1 : Aligned Second Sequence is '{0}'.", expectedSequence2));
}
/// <summary>
/// Construct an aligner and run the alignment.
/// </summary>
/// <param name="sequences">input sequences</param>
/// <param name="kmerLength">positive integer of kmer length</param>
/// <param name="distanceFunctionName">enum: distance function name</param>
/// <param name="hierarchicalClusteringMethodName">enum: cluster update method</param>
/// <param name="profileAlignerMethodName">enum: profile-profile aligner name</param>
/// <param name="profileFunctionName">enum: profile-profile distance function</param>
/// <param name="similarityMatrix">similarity matrix</param>
/// <param name="gapOpenPenalty">negative gapOpenPenalty</param>
/// <param name="gapExtendPenalty">negative gapExtendPenalty</param>
/// <param name="numberOfPartitions">the number of partitions in dynamic programming</param>
/// <param name="degreeOfParallelism">degree of parallelism option for parallel extension</param>
public PAMSAMMultipleSequenceAligner(
IList<ISequence> sequences,
int kmerLength,
DistanceFunctionTypes distanceFunctionName,
UpdateDistanceMethodsTypes hierarchicalClusteringMethodName,
ProfileAlignerNames profileAlignerMethodName,
ProfileScoreFunctionNames profileFunctionName,
SimilarityMatrix similarityMatrix,
int gapOpenPenalty,
int gapExtendPenalty,
int numberOfPartitions,
int degreeOfParallelism)
{
AlignmentScoreC = float.MinValue;
AlignmentScoreB = float.MinValue;
AlignmentScoreA = float.MinValue;
AlignmentScore = float.MinValue;
StartLog();
if (null == sequences)
{
throw new ArgumentNullException("sequences");
}
if (sequences.Count == 0)
{
throw new ArgumentException("Empty input sequences");
}
// Set parallel extension option
if (degreeOfParallelism <= 0)
{
throw new ArgumentException("Invalid parallel degree parameter");
}
this.degreeOfParallelism = degreeOfParallelism;
ParallelOption = new ParallelOptions { MaxDegreeOfParallelism = degreeOfParallelism };
if (numberOfPartitions <= 0)
{
throw new ArgumentException("Invalid number of partition parameter");
}
this.numberOfPartitions = numberOfPartitions;
// Assign the alphabet
SetAlphabet(sequences, similarityMatrix, false);
// Initialize parameters
KmerLength = kmerLength;
DistanceFunctionName = distanceFunctionName;
HierarchicalClusteringMethodName = hierarchicalClusteringMethodName;
ProfileAlignerName = profileAlignerMethodName;
ProfileProfileFunctionName = profileFunctionName;
SimilarityMatrix = similarityMatrix;
GapOpenCost = gapOpenPenalty;
GapExtensionCost = gapExtendPenalty;
MsaUtils.SetProfileItemSets(this.alphabet);
ReportLog("Start Aligning");
// Work...
DoAlignment(sequences);
}
/// <summary>
/// Log of Weighted inner-product by exponential of shifted similarity matrix
/// </summary>
/// <param name="similarityMatrix">similarity matrix</param>
/// <param name="profileIndexA">the first profile vector (normalized)</param>
/// <param name="profileIndexB">the second profile vector (normalized)</param>
protected float LogExponentialInnerProductShiftedFast(SimilarityMatrix similarityMatrix, int profileIndexA, int profileIndexB)
{
float[] profileA = _profileAlignmentA.ProfilesMatrix[profileIndexA];
float[] profileB = _profileAlignmentB.ProfilesMatrix[profileIndexB];
if (profileA.Length != profileB.Length)
{
throw new ArgumentException("Unequal length profiles");
}
int dimension = profileA.Length - 1;
float result = 0;
_indexA = MsaUtils.CreateIndexArray(dimension);
_indexB = MsaUtils.CreateIndexArray(dimension);
//MsaUtils.QuickSort(a, aIndex, 0, a.Length - 1);
MsaUtils.QuickSort(profileA, _indexA, 0, dimension - 1);
MsaUtils.QuickSort(profileB, _indexB, 0, dimension - 1);
for (int i = 0; i < dimension; ++i)
{
if (profileA[_indexA[i]] == 0)
{
break;
}
for (int j = 0; j < dimension; ++j)
{
if (profileB[_indexB[j]] == 0)
{
break;
}
result += profileB[j] * profileA[i] * (float)Math.Pow(2, similarityMatrix[i, j] + 0.5);
}
}
result *= (1 - profileA[dimension]) * (1 - profileB[dimension]);
return (float)Math.Log(result, 2);
}
/// <summary>
/// Log of Weighted inner-product by exponential of shifted similarity matrix
/// </summary>
/// <param name="similarityMatrix">similarity matrix</param>
/// <param name="profileIndexA">the first profile vector (normalized)</param>
/// <param name="profileIndexB">the second profile vector (normalized)</param>
protected float LogExponentialInnerProductShifted(SimilarityMatrix similarityMatrix, int profileIndexA, int profileIndexB)
{
float[] profileA = _profileAlignmentA.ProfilesMatrix[profileIndexA];
float[] profileB = _profileAlignmentB.ProfilesMatrix[profileIndexB];
if (profileA.Length != profileB.Length)
{
throw new ArgumentException("Unequal length profiles");
}
int dimension = profileA.Length - 1;
float result = 0;
for (int i = 0; i < dimension; ++i)
{
for (int j = 0; j < dimension; ++j)
{
result += profileA[i] * profileB[j] * (float)Math.Pow(2, similarityMatrix[i, j] + 0.5);
}
}
result *= (profileA[dimension]) * (1 - profileB[dimension]);
return (float)Math.Log(result, 2);
}
/// <summary>
/// Correlation of observation vectors
/// </summary>
/// <param name="similarityMatrix">similarity matrix</param>
/// <param name="profileIndexA">the first profile vector (normalized)</param>
/// <param name="profileIndexB">the second profile vector (normalized)</param>
protected float PearsonCorrelation(SimilarityMatrix similarityMatrix, int profileIndexA, int profileIndexB)
{
float[] profileA = _profileAlignmentA.ProfilesMatrix[profileIndexA];
float[] profileB = _profileAlignmentB.ProfilesMatrix[profileIndexB];
if (profileA.Length != profileB.Length)
{
throw new ArgumentException("Unequal length profiles");
}
return MsaUtils.Correlation(profileA, profileB);
}
/// <summary>
/// Weighted inner-profuct by shifted similarity matrix
/// </summary>
/// <param name="similarityMatrix">similarity matrix</param>
/// <param name="profileIndexA">the first profile vector (normalized)</param>
/// <param name="profileIndexB">the second profile vector (normalized)</param>
protected float WeightedInnerProductShifted(SimilarityMatrix similarityMatrix, int profileIndexA, int profileIndexB)
{
float[] profileA = _profileAlignmentA.ProfilesMatrix[profileIndexA];
float[] profileB = _profileAlignmentB.ProfilesMatrix[profileIndexB];
if (profileA.Length != profileB.Length)
{
throw new ArgumentException("Unequal length profiles");
}
int dimension = profileA.Length - 1;
float[] cachedW = new float[dimension];
for (int i = 0; i < dimension; ++i)
{
for (int j = 0; j < dimension; ++j)
{
cachedW[i] += (similarityMatrix[i, j] + (float)0.5) * profileB[j];
}
}
float result = 0;
for (int i = 0; i < dimension; ++i)
{
result += profileA[i] * cachedW[i];
}
result *= (1 - profileA[dimension]) * (1 - profileB[dimension]);
return result;
}
/// <summary>
/// This method assigns the alphabet from the input sequences
/// </summary>
/// <param name="sequences">Input sequences</param>
/// <param name="similarityMatrix">Matrix to use for similarity comparisons</param>
/// <param name="fixSimilarityMatrixErrors">True to fix any similarity matrix issue related to the alphabet.</param>
private void SetAlphabet(IList<ISequence> sequences, SimilarityMatrix similarityMatrix, bool fixSimilarityMatrixErrors)
{
if (sequences.Count == 0)
{
throw new ArgumentException("Empty input sequences");
}
// Validate data type
this.alphabet = Alphabets.GetAmbiguousAlphabet(sequences[0].Alphabet);
Parallel.For(1, sequences.Count, ParallelOption, i =>
{
if (!Alphabets.CheckIsFromSameBase(sequences[i].Alphabet, this.alphabet))
{
throw new ArgumentException("Inconsistent sequence alphabet");
}
});
SimilarityMatrix bestSimilarityMatrix = null;
if (this.alphabet is DnaAlphabet)
{
bestSimilarityMatrix = new SimilarityMatrix(SimilarityMatrix.StandardSimilarityMatrix.AmbiguousDna);
}
else if (this.alphabet is RnaAlphabet)
{
bestSimilarityMatrix = new SimilarityMatrix(SimilarityMatrix.StandardSimilarityMatrix.AmbiguousRna);
}
else if (this.alphabet is ProteinAlphabet)
{
bestSimilarityMatrix = new SimilarityMatrix(SimilarityMatrix.StandardSimilarityMatrix.Blosum50);
}
// Check or assign the similarity matrix.
if (similarityMatrix == null)
{
SimilarityMatrix = bestSimilarityMatrix;
if (SimilarityMatrix == null)
throw new ArgumentException("Unknown alphabet - could not choose SimilarityMatrix.");
}
else
{
var similarityMatrixDNA = new List<String> { "AmbiguousDNA" };
var similarityMatrixRNA = new List<String> { "AmbiguousRNA" };
var similarityMatrixProtein = new List<String> { "BLOSUM45", "BLOSUM50", "BLOSUM62", "BLOSUM80", "BLOSUM90", "PAM250", "PAM30", "PAM70" };
if (this.alphabet is DnaAlphabet)
{
if (!similarityMatrixDNA.Contains(similarityMatrix.Name))
{
if (fixSimilarityMatrixErrors)
SimilarityMatrix = bestSimilarityMatrix;
else
throw new ArgumentException("Inappropriate Similarity Matrix for DNA.");
}
}
else if (this.alphabet is ProteinAlphabet)
{
if (!similarityMatrixProtein.Contains(similarityMatrix.Name))
{
if (fixSimilarityMatrixErrors)
SimilarityMatrix = bestSimilarityMatrix;
else
throw new ArgumentException("Inappropriate Similarity Matrix for Protein.");
}
}
else if (this.alphabet is RnaAlphabet)
{
if (!similarityMatrixRNA.Contains(similarityMatrix.Name))
{
if (fixSimilarityMatrixErrors)
SimilarityMatrix = bestSimilarityMatrix;
else
throw new ArgumentException("Inappropriate Similarity Matrix for RNA.");
}
}
else
{
throw new ArgumentException("Invalid alphabet");
}
}
}
/// <summary>
/// Jensen-Shannon divergence of observation vectors
/// </summary>
/// <param name="similarityMatrix">similarity matrix</param>
/// <param name="profileIndexA">the first profile vector (normalized)</param>
/// <param name="profileIndexB">the second profile vector (normalized)</param>
protected float JensenShannonDivergence(SimilarityMatrix similarityMatrix, int profileIndexA, int profileIndexB)
{
return 1 - MsaUtils.JensenShannonDivergence(_profileAlignmentA.ProfilesMatrix[profileIndexA], _profileAlignmentB.ProfilesMatrix[profileIndexB]);
}
private void ValidateSmithWatermanAlignment(string nodeName, bool isTextFile,
SequenceCaseType caseType, AlignParameters additionalParameter,
AlignmentType alignType,
SimilarityMatrixParameters similarityMatrixParam)
{
Sequence aInput, bInput;
IAlphabet alphabet =
Utility.GetAlphabet(this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.AlphabetNameNode));
if (isTextFile)
{
// Read the xml file for getting both the files for aligning.
string filePath1 = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.FilePathNode1);
string filePath2 = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.FilePathNode2);
// Parse the files and get the sequence.
ISequence originalSequence1 = null;
ISequence originalSequence2 = null;
var parseObjectForFile1 = new FastAParser { Alphabet = alphabet };
originalSequence1 = parseObjectForFile1.Parse(filePath1).ElementAt(0);
originalSequence2 = parseObjectForFile1.Parse(filePath2).ElementAt(0);
// Create input sequence for sequence string in different cases.
GetSequenceWithCaseType(originalSequence1.ConvertToString(), originalSequence2.ConvertToString(),
alphabet, caseType, out aInput, out bInput);
}
else
{
string originalSequence1 = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.SequenceNode1);
string originalSequence2 = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.SequenceNode2);
// Create input sequence for sequence string in different cases.
GetSequenceWithCaseType(
originalSequence1,
originalSequence2,
alphabet,
caseType,
out aInput,
out bInput);
}
ApplicationLog.WriteLine(string.Format("SmithWatermanAligner P2 : First sequence used is '{0}'.",
aInput.ConvertToString()));
ApplicationLog.WriteLine(string.Format("SmithWatermanAligner P2 : Second sequence used is '{0}'.",
bInput.ConvertToString()));
// Create similarity matrix object for a given file.
string blosumFilePath = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.BlosumFilePathNode);
SimilarityMatrix sm;
switch (similarityMatrixParam)
{
case SimilarityMatrixParameters.TextReader:
using (TextReader reader = new StreamReader(blosumFilePath))
sm = new SimilarityMatrix(reader);
break;
case SimilarityMatrixParameters.DiagonalMatrix:
string matchValue = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.MatchScoreNode);
string misMatchValue = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.MisMatchScoreNode);
sm = new DiagonalSimilarityMatrix(int.Parse(matchValue, null),
int.Parse(misMatchValue, null));
break;
default:
sm = new SimilarityMatrix(new StreamReader(blosumFilePath));
break;
}
int gapOpenCost = int.Parse(this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.GapOpenCostNode), null);
int gapExtensionCost = int.Parse(this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.GapExtensionCostNode),
null);
// Create SmithWatermanAligner instance and set its values.
var smithWatermanObj = new SmithWatermanAligner();
if (additionalParameter != AlignParameters.AllParam)
{
smithWatermanObj.SimilarityMatrix = sm;
smithWatermanObj.GapOpenCost = gapOpenCost;
smithWatermanObj.GapExtensionCost = gapExtensionCost;
}
IList<IPairwiseSequenceAlignment> result = null;
// Align the input sequences.
switch (additionalParameter)
{
case AlignParameters.AlignList:
switch (alignType)
{
case AlignmentType.Align:
result = smithWatermanObj.Align(new List<ISequence> {aInput, bInput});
break;
default:
result = smithWatermanObj.AlignSimple(new List<ISequence> {aInput, bInput});
break;
}
break;
case AlignParameters.AlignTwo:
switch (alignType)
{
case AlignmentType.Align:
result = smithWatermanObj.Align(aInput, bInput);
break;
default:
result = smithWatermanObj.AlignSimple(aInput, bInput);
break;
}
break;
case AlignParameters.AllParam:
switch (alignType)
{
case AlignmentType.Align:
result = smithWatermanObj.Align(sm, gapOpenCost,
gapExtensionCost, aInput, bInput);
break;
default:
result = smithWatermanObj.AlignSimple(sm, gapOpenCost, aInput, bInput);
break;
}
break;
default:
break;
}
// Get the expected sequence and scorde from xml config.
string expectedSequence1, expectedSequence2, expectedScore;
switch (alignType)
{
case AlignmentType.Align:
expectedScore = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.ExpectedGapExtensionScoreNode);
switch (caseType)
{
case SequenceCaseType.LowerCase:
expectedSequence1 = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants
.ExpectedGapExtensionSequence1InLower);
expectedSequence2 = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants
.ExpectedGapExtensionSequence2InLower);
break;
default:
expectedSequence1 = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants
.ExpectedGapExtensionSequence1Node);
expectedSequence2 = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants
.ExpectedGapExtensionSequence2Node);
break;
}
break;
default:
expectedScore = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.ExpectedScoreNode);
switch (caseType)
{
case SequenceCaseType.LowerCase:
expectedSequence1 = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.ExpectedSequence1inLowerNode);
expectedSequence2 = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.ExpectedSequence2inLowerNode);
break;
case SequenceCaseType.LowerUpperCase:
expectedSequence1 = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.ExpectedSequence1inLowerNode);
expectedSequence2 = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.ExpectedSequenceNode2);
break;
default:
expectedSequence1 = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.ExpectedSequenceNode1);
expectedSequence2 = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.ExpectedSequenceNode2);
break;
}
break;
}
// Match the alignment result with expected result.
IList<IPairwiseSequenceAlignment> expectedOutput = new List<IPairwiseSequenceAlignment>();
IPairwiseSequenceAlignment align = new PairwiseSequenceAlignment();
var alignedSeq = new PairwiseAlignedSequence
{
FirstSequence = new Sequence(alphabet, expectedSequence1),
SecondSequence = new Sequence(alphabet, expectedSequence2),
Score = Convert.ToInt32(expectedScore, null),
FirstOffset = Int32.MinValue,
SecondOffset = Int32.MinValue,
};
align.PairwiseAlignedSequences.Add(alignedSeq);
expectedOutput.Add(align);
ApplicationLog.WriteLine(string.Format(null, "SmithWatermanAligner P2 : Final Score '{0}'.", expectedScore));
ApplicationLog.WriteLine(string.Format(null, "SmithWatermanAligner P2 : Aligned First Sequence is '{0}'.",
expectedSequence1));
ApplicationLog.WriteLine(string.Format(null, "SmithWatermanAligner P2 : Aligned Second Sequence is '{0}'.",
expectedSequence2));
Assert.IsTrue(CompareAlignment(result, expectedOutput));
}
/// <summary>
/// Constructor for all the pairwise aligner (NeedlemanWunsch, SmithWaterman, Overlap).
/// Sets default similarity matrix and gap penalties.
/// Users will typically reset these using parameters specific to their particular sequences and needs.
///
/// This constructor is for non-parallel version.
/// </summary>
protected DynamicProgrammingProfileAlignerSerial(
SimilarityMatrix similarityMatrix,
ProfileScoreFunctionNames profileScoreFunctionName,
int gapOpenPenalty,
int gapExtensionPenalty)
: this(similarityMatrix, profileScoreFunctionName, gapOpenPenalty, gapExtensionPenalty, 1)
{
}
private void InValidateSmithWatermanAlignmentWithInvalidSimilarityMatrix(string nodeName,
bool isTextFile,
SimilarityMatrixInvalidTypes
invalidType,
AlignParameters additionalParameter,
AlignmentType alignType)
{
Sequence aInput = null;
Sequence bInput = null;
ISequence inputSequence1;
ISequence inputSequence2;
IAlphabet alphabet = Utility.GetAlphabet(this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.AlphabetNameNode));
if (isTextFile)
{
// Read the xml file for getting both the files for aligning.
string firstInputFilepath = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.FilePathNode1);
string secondInputFilepath = this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.FilePathNode2);
// Parse the files and get the sequence.
var parseObjectForFile1 = new FastAParser { Alphabet = alphabet };
inputSequence1 = parseObjectForFile1.Parse(firstInputFilepath).ElementAt(0);
inputSequence2 = parseObjectForFile1.Parse(secondInputFilepath).ElementAt(0);
// Create input sequence for sequence string in different cases.
GetSequenceWithCaseType(new string(inputSequence1.Select(a => (char) a).ToArray()),
new string(inputSequence2.Select(a => (char) a).ToArray()),
alphabet, SequenceCaseType.LowerCase, out aInput, out bInput);
}
else
{
string firstInputSequence = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.SequenceNode1);
string secondInputSequence = this.utilityObj.xmlUtil.GetTextValue(nodeName, Constants.SequenceNode2);
// Create input sequence for sequence string in different cases.
GetSequenceWithCaseType(firstInputSequence, secondInputSequence,
alphabet, SequenceCaseType.LowerCase, out aInput, out bInput);
}
ApplicationLog.WriteLine(string.Concat(
"SmithWatermanAligner P2 : First sequence used is '{0}'.",
new string(aInput.Select(a => (char) a).ToArray())));
ApplicationLog.WriteLine(string.Concat(
"SmithWatermanAligner P2 : Second sequence used is '{0}'.",
new string(bInput.Select(a => (char) a).ToArray())));
// Create similarity matrix object for a invalid file.
string blosumFilePath = this.GetSimilarityMatrixFileWithInvalidType(nodeName, invalidType);
Exception actualExpection = null;
// For invalid similarity matrix data format; exception will be thrown while instantiating
SimilarityMatrix sm = null;
try
{
if (invalidType != SimilarityMatrixInvalidTypes.NullSimilarityMatrix)
{
sm = new SimilarityMatrix(new StreamReader(blosumFilePath));
}
}
catch (InvalidDataException ex)
{
actualExpection = ex;
}
// For non matching similarity matrix exception will be thrown while alignment
if (actualExpection == null)
{
int gapOpenCost = int.Parse(this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.GapOpenCostNode), null);
int gapExtensionCost = int.Parse(this.utilityObj.xmlUtil.GetTextValue(nodeName,
Constants.GapExtensionCostNode), null);
// Create SmithWatermanAligner instance and set its values.
var smithWatermanObj = new SmithWatermanAligner();
if (additionalParameter != AlignParameters.AllParam)
{
smithWatermanObj.SimilarityMatrix = sm;
smithWatermanObj.GapOpenCost = gapOpenCost;
smithWatermanObj.GapExtensionCost = gapExtensionCost;
}
// Align the input sequences and catch the exception.
switch (additionalParameter)
{
case AlignParameters.AlignList:
switch (alignType)
{
case AlignmentType.Align:
try
{
smithWatermanObj.Align(new List<ISequence> {aInput, bInput});
}
catch (ArgumentException ex)
{
actualExpection = ex;
}
break;
default:
try
{
smithWatermanObj.AlignSimple(new List<ISequence> {aInput, bInput});
}
catch (ArgumentException ex)
{
actualExpection = ex;
}
break;
}
break;
case AlignParameters.AlignTwo:
switch (alignType)
{
case AlignmentType.Align:
try
{
smithWatermanObj.Align(aInput, bInput);
}
catch (ArgumentException ex)
{
actualExpection = ex;
}
break;
default:
try
{
smithWatermanObj.AlignSimple(aInput, bInput);
}
catch (ArgumentException ex)
{
actualExpection = ex;
}
break;
}
break;
case AlignParameters.AllParam:
switch (alignType)
{
case AlignmentType.Align:
try
{
smithWatermanObj.Align(sm, gapOpenCost, gapExtensionCost,
aInput, bInput);
}
catch (ArgumentException ex)
{
actualExpection = ex;
}
break;
default:
try
{
smithWatermanObj.AlignSimple(sm, gapOpenCost, aInput, bInput);
}
catch (ArgumentException ex)
{
actualExpection = ex;
}
break;
}
break;
default:
break;
}
}
// Validate that expected exception is thrown using error message.
string expectedErrorMessage = this.GetExpectedErrorMeesageWithInvalidSimilarityMatrixType(nodeName,
invalidType);
Assert.AreEqual(expectedErrorMessage, actualExpection.Message);
ApplicationLog.WriteLine(string.Concat(
"SmithWatermanAligner P2 : Expected Error message is thrown ", expectedErrorMessage));
}
/// <summary>
/// Constructor for all the pairwise aligner (NeedlemanWunsch, SmithWaterman, Overlap).
/// Sets default similarity matrix and gap penalties.
/// Users will typically reset these using parameters specific to their particular sequences and needs.
/// </summary>
protected DynamicProgrammingProfileAlignerSerial(
SimilarityMatrix similarityMatrix,
ProfileScoreFunctionNames profileScoreFunctionName,
int gapOpenPenalty,
int gapExtensionPenalty,
int numberOfCores)
{
// Set default similarity matrix and gap penalty.
// User will typically choose their own parameters, these defaults are reasonable for many cases.
// Molecule type is set to protein, since this will also work for DNA and RNA in the
// special case of a diagonal similarity matrix.
_similarityMatrix = similarityMatrix;
_gapOpenPenalty = gapOpenPenalty;
_gapExtensionPenalty = gapExtensionPenalty;
switch (profileScoreFunctionName)
{
case (ProfileScoreFunctionNames.InnerProduct):
_profileProfileScoreFunction = new ProfileScoreFunctionSelector(InnerProduct);
break;
case (ProfileScoreFunctionNames.WeightedInnerProduct):
_profileProfileScoreFunction = new ProfileScoreFunctionSelector(WeightedInnerProduct);
break;
case (ProfileScoreFunctionNames.WeightedInnerProductShifted):
_profileProfileScoreFunction = new ProfileScoreFunctionSelector(WeightedInnerProductShifted);
break;
case (ProfileScoreFunctionNames.InnerProductFast):
_profileProfileScoreFunction = new ProfileScoreFunctionSelector(InnerProductFast);
break;
case (ProfileScoreFunctionNames.WeightedInnerProductFast):
_profileProfileScoreFunction = new ProfileScoreFunctionSelector(WeightedInnerProductFast);
break;
case (ProfileScoreFunctionNames.WeightedInnerProductShiftedFast):
_profileProfileScoreFunction = new ProfileScoreFunctionSelector(WeightedInnerProductShiftedFast);
break;
case (ProfileScoreFunctionNames.PearsonCorrelation):
_profileProfileScoreFunction = new ProfileScoreFunctionSelector(PearsonCorrelation);
break;
case (ProfileScoreFunctionNames.WeightedEuclideanDistance):
_profileProfileScoreFunction = new ProfileScoreFunctionSelector(WeightedEuclideanDistance);
break;
case (ProfileScoreFunctionNames.LogExponentialInnerProduct):
_profileProfileScoreFunction = new ProfileScoreFunctionSelector(LogExponentialInnerProduct);
break;
case (ProfileScoreFunctionNames.LogExponentialInnerProductShifted):
_profileProfileScoreFunction = new ProfileScoreFunctionSelector(LogExponentialInnerProductShifted);
break;
case (ProfileScoreFunctionNames.WeightedEuclideanDistanceFast):
_profileProfileScoreFunction = new ProfileScoreFunctionSelector(WeightedEuclideanDistanceFast);
break;
case (ProfileScoreFunctionNames.LogExponentialInnerProductFast):
_profileProfileScoreFunction = new ProfileScoreFunctionSelector(LogExponentialInnerProductFast);
break;
case (ProfileScoreFunctionNames.LogExponentialInnerProductShiftedFast):
_profileProfileScoreFunction = new ProfileScoreFunctionSelector(LogExponentialInnerProductShiftedFast);
break;
case (ProfileScoreFunctionNames.SymmetrizedEntropy):
_profileProfileScoreFunction = new ProfileScoreFunctionSelector(SymmetrizedEntropy);
break;
case (ProfileScoreFunctionNames.JensenShannonDivergence):
_profileProfileScoreFunction = new ProfileScoreFunctionSelector(JensenShannonDivergence);
break;
case (ProfileScoreFunctionNames.WeightedInnerProductCached):
_profileProfileScoreFunction = new ProfileScoreFunctionSelector(WeightedInnerProductCached);
_cachingFunction = new CachingFunctionSelector(CachingWeightedInnerProduct);
_doCaching = true;
break;
default:
throw new Exception("Invalid profile function name");
}
if (numberOfCores <= 0)
{
throw new ArgumentException("Invalid number of cores parameter");
}
_numberOfPartitions = numberOfCores;
}
/// <summary>
/// Inializes a new alignment job
/// </summary>
/// <param name="similarityMatrix"></param>
/// <param name="gapOpenCost"></param>
/// <param name="gapExtensionCost"></param>
/// <param name="aInput"></param>
/// <param name="bInput"></param>
protected PairwiseOverlapSimpleAlignmentJob(SimilarityMatrix similarityMatrix, int gapOpenCost, int gapExtensionCost, ISequence aInput, ISequence bInput)
: base(similarityMatrix, gapOpenCost, gapExtensionCost, aInput, bInput) { }
/// <summary>
/// Pairwise alignment of two sequences using an affine gap penalty. The various algorithms in derived classes (NeedlemanWunsch,
/// SmithWaterman, and PairwiseOverlap) all use this general engine for alignment with an affine gap penalty.
/// </summary>
/// <param name="similarityMatrix">Scoring matrix.</param>
/// <param name="gapOpenPenalty">Gap open penalty (by convention, use a negative number for this.)</param>
/// <param name="gapExtensionPenalty">Gap extension penalty (by convention, use a negative number for this.)</param>
/// <param name="profileAlignmentA">First input profileAlignment</param>
/// <param name="profileAlignmentB">Second input profileAlignment</param>
public IProfileAlignment Align(
SimilarityMatrix similarityMatrix,
int gapOpenPenalty,
int gapExtensionPenalty,
IProfileAlignment profileAlignmentA,
IProfileAlignment profileAlignmentB)
{
_profileAlignmentA = profileAlignmentA;
_profileAlignmentB = profileAlignmentB;
ResetSpecificAlgorithmMemberVariables();
// Set Gap Penalty and Similarity Matrix
GapOpenCost = gapOpenPenalty;
GapExtensionCost = gapExtensionPenalty;
SimilarityMatrix = similarityMatrix;
ValidateAlignInput(profileAlignmentA, profileAlignmentB); // throws exception if input not valid
// Convert input strings to 0-based int arrays using similarity matrix mapping
_a = MsaUtils.CreateIndexArray(profileAlignmentA.ProfilesMatrix.RowSize);
_b = MsaUtils.CreateIndexArray(profileAlignmentB.ProfilesMatrix.RowSize);
if (_doCaching)
{
_cachingFunction(similarityMatrix, _profileAlignmentA, _profileAlignmentB);
}
// Sort profileA
_indexAs = CachingIndex(_profileAlignmentA);
FillMatrixAffine();
//DumpF(); // Writes matrix to application log, used for development and testing
//DumpAffine(); // Writes matrix to application log in great detail. Useful only for small cases.
float optScore = Traceback(out _alignedA, out _alignedB);
#region Convert aligned sequences back to Sequence objects, load output SequenceAlignment object
IProfileAlignment results = null;
//AddSimpleConsensusToResult(results);
if (PAMSAMMultipleSequenceAligner.UseWeights)
{
results = ProfileAlignment.GenerateProfileAlignment(_profileAlignmentA, _profileAlignmentB, _alignedA, _alignedB, _gapCode, _weights);
}
else
{
results = ProfileAlignment.GenerateProfileAlignment(_profileAlignmentA, _profileAlignmentB, _alignedA, _alignedB, _gapCode);
}
results.Score = optScore;
#endregion
return results;
}
/// <summary>
/// User will typically choose their own parameters, these defaults are reasonable for many cases.
/// </summary>
private void SetDefaults()
{
_nucmerAligner = new ModifiedSmithWaterman();
// Set the default Similarity Matrix
SimilarityMatrix = new SimilarityMatrix(
SimilarityMatrix.StandardSimilarityMatrix.DiagonalScoreMatrix);
// Set the defaults
GapOpenCost = DefaultGapOpenCost;
GapExtensionCost = DefaultGapExtensionCost;
LengthOfMUM = DefaultLengthOfMUM;
// Set the ClusterBuilder properties to defaults
FixedSeparation = ClusterBuilder.DefaultFixedSeparation;
MaximumSeparation = ClusterBuilder.DefaultMaximumSeparation;
MinimumScore = ClusterBuilder.DefaultMinimumScore;
SeparationFactor = ClusterBuilder.DefaultSeparationFactor;
BreakLength = ModifiedSmithWaterman.DefaultBreakLength;
ScoreMethod = ClusterScoreMethod.MatchLength;
}
/// <summary>
/// Pairwise alignment of two sequences using a linear gap penalty. The various algorithms in derived classes (NeedlemanWunsch,
/// SmithWaterman, and PairwiseOverlap) all use this general engine for alignment with a linear gap penalty.
/// </summary>
/// <param name="localSimilarityMatrix">Scoring matrix.</param>
/// <param name="gapPenalty">Gap penalty (by convention, use a negative number for this.).</param>
/// <param name="inputA">First input sequence.</param>
/// <param name="inputB">Second input sequence.</param>
/// <returns>A list of sequence alignments.</returns>
public IList<IPairwiseSequenceAlignment> AlignSimple(SimilarityMatrix localSimilarityMatrix, int gapPenalty, ISequence inputA, ISequence inputB)
{
this.SimilarityMatrix = localSimilarityMatrix;
this.GapOpenCost = gapPenalty;
return DoAlign(inputA, inputB,false);
}
