/// <summary>
/// Initializes a new instance of the OverlapDeNovoAssembler class.
/// Sets default threshold values, pairwise aligner, consensusResolver.
/// Users will typically reset these using parameters
/// specific to their particular sequences and needs.
/// </summary>
public OverlapDeNovoAssembler()
{
// The following definitions give the default values
// for different parameters used for initialization
// Note: Each of the following constructor calls,
// in turn might set default values for its required parameters.
// By default, set merge threshold to 3
MergeThreshold = 3;
// By default, use PairwiseOverlapAligner
OverlapAlgorithm = new PairwiseOverlapAligner();
// By default, value of AssumeStandardOrientation is set as true
AssumeStandardOrientation = true;
}
public void PairwiseOverlapProteinSeqSimpleGap()
{
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;
IList<IPairwiseSequenceAlignment> result = overlap.AlignSimple(sequence1, sequence2);
ApplicationLog.WriteLine(string.Format((IFormatProvider)null,
"{0}, Simple; Matrix {1}; GapOpenCost {2}", overlap.Name, overlap.SimilarityMatrix.Name, overlap.GapOpenCost));
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>
/// 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));
}
public void PairwiseOverlapMultipleAlignments()
{
Sequence sequence1 = new Sequence(Alphabets.DNA, "CCCAACCC");
Sequence sequence2 = new Sequence(Alphabets.DNA, "CCC");
SimilarityMatrix sm = new DiagonalSimilarityMatrix(5, -20);
int gapPenalty = -10;
PairwiseOverlapAligner overlap = new PairwiseOverlapAligner();
overlap.SimilarityMatrix = sm;
overlap.GapOpenCost = gapPenalty;
IList<IPairwiseSequenceAlignment> result = overlap.AlignSimple(sequence1, sequence2);
ApplicationLog.WriteLine(string.Format((IFormatProvider)null,
"{0}, Simple; Matrix {1}; GapOpenCost {2}", overlap.Name, overlap.SimilarityMatrix.Name, overlap.GapOpenCost));
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();
// First alignment
PairwiseAlignedSequence alignedSeq = new PairwiseAlignedSequence();
alignedSeq.FirstSequence = new Sequence(Alphabets.DNA, "CCC");
alignedSeq.SecondSequence = new Sequence(Alphabets.DNA, "CCC");
alignedSeq.Consensus = new Sequence(Alphabets.DNA, "CCC");
alignedSeq.Score = 15;
alignedSeq.FirstOffset = 0;
alignedSeq.SecondOffset = 0;
align.PairwiseAlignedSequences.Add(alignedSeq);
// Second alignment
alignedSeq = new PairwiseAlignedSequence();
alignedSeq.FirstSequence = new Sequence(Alphabets.DNA, "CCC");
alignedSeq.SecondSequence = new Sequence(Alphabets.DNA, "CCC");
alignedSeq.Consensus = new Sequence(Alphabets.DNA, "CCC");
alignedSeq.Score = 15;
alignedSeq.FirstOffset = 0;
alignedSeq.SecondOffset = 5;
align.PairwiseAlignedSequences.Add(alignedSeq);
expectedOutput.Add(align);
Assert.IsTrue(CompareAlignment(result, expectedOutput));
}
public void PairwiseOverlapProteinSeqWithZeroOverlap()
{
Sequence sequence1 = new Sequence(Alphabets.Protein, "ACDEF");
Sequence sequence2 = new Sequence(Alphabets.Protein, "TUVWY");
SimilarityMatrix sm = new DiagonalSimilarityMatrix(5, -5);
int gapPenalty = -10;
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}, Simple; Matrix {1}; GapOpenCost {2}", overlap.Name, overlap.SimilarityMatrix.Name, overlap.GapOpenCost));
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>();
Assert.IsTrue(CompareAlignment(result, expectedOutput));
}
/// <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="invalidType">Invalid 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>
private void InValidatePairwiseOverlapAlignmentWithInvalidSimilarityMatrix(string nodeName,
bool isTextFile,
SimilarityMatrixInvalidTypes
invalidType,
AlignParameters additionalParameter,
AlignmentType alignType)
{
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 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 parser1 = new FastAParser();
ISequence inputSequence1 = parser1.Parse(firstInputFilepath).ElementAt(0);
ISequence inputSequence2 = parser1.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);
}
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 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 PairwiseOverlapAligner instance and set its values.
var pairwiseOverlapObj = new PairwiseOverlapAligner();
if (additionalParameter != AlignParameters.AllParam)
{
pairwiseOverlapObj.SimilarityMatrix = sm;
pairwiseOverlapObj.GapOpenCost = gapOpenCost;
pairwiseOverlapObj.GapExtensionCost = gapExtensionCost;
}
// Align the input sequences and catch the exception.
switch (additionalParameter)
{
case AlignParameters.AlignList:
var sequences = new List<ISequence>();
sequences.Add(aInput);
sequences.Add(bInput);
switch (alignType)
{
case AlignmentType.Align:
try
{
pairwiseOverlapObj.Align(sequences);
}
catch (ArgumentException ex)
{
actualExpection = ex;
}
break;
default:
try
{
pairwiseOverlapObj.AlignSimple(sequences);
}
catch (ArgumentException ex)
{
actualExpection = ex;
}
break;
}
break;
case AlignParameters.AlignTwo:
switch (alignType)
{
case AlignmentType.Align:
try
{
pairwiseOverlapObj.Align(aInput, bInput);
}
catch (ArgumentException ex)
{
actualExpection = ex;
}
break;
default:
try
{
pairwiseOverlapObj.AlignSimple(aInput, bInput);
}
catch (ArgumentException ex)
{
actualExpection = ex;
}
break;
}
break;
case AlignParameters.AllParam:
switch (alignType)
{
case AlignmentType.Align:
try
{
pairwiseOverlapObj.Align(sm, gapOpenCost, gapExtensionCost,
aInput, bInput);
}
catch (ArgumentException ex)
{
actualExpection = ex;
}
break;
default:
try
{
pairwiseOverlapObj.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("PairwiseOverlapAligner P2 : Expected Error message is thrown ", expectedErrorMessage);
}
/// <summary>
/// InValidates PairwiseOverlapAlignment with invalid sequence.
/// </summary>
/// <param name="nodeName"></param>
/// <param name="isTextFile">Is text file an input.</param>
/// <param name="invalidSequenceType"></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="sequenceType"></param>
private void InValidatePairwiseOverlapAlignmentWithInvalidSequence(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;
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.
var parser = new FastAParser();
aInput = new Sequence(alphabet, new string(parser.Parse(filepath).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 (null == actualException)
{
Sequence 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 PairwiseOverlapAligner instance and set its values.
var pairwiseOverlapObj = new PairwiseOverlapAligner();
if (additionalParameter != AlignParameters.AllParam)
{
pairwiseOverlapObj.SimilarityMatrix = sm;
pairwiseOverlapObj.GapOpenCost = gapOpenCost;
pairwiseOverlapObj.GapExtensionCost = gapExtensionCost;
}
// Align the input sequences and catch the exception.
switch (additionalParameter)
{
case AlignParameters.AlignList:
var sequences = new List<ISequence> {aInput, bInput};
switch (alignType)
{
case AlignmentType.Align:
try
{
pairwiseOverlapObj.Align(sequences);
}
catch (ArgumentException ex)
{
actualException = ex;
}
break;
default:
try
{
pairwiseOverlapObj.AlignSimple(sequences);
}
catch (ArgumentException ex)
{
actualException = ex;
}
break;
}
break;
case AlignParameters.AlignTwo:
switch (alignType)
{
case AlignmentType.Align:
try
{
pairwiseOverlapObj.Align(aInput, bInput);
}
catch (ArgumentException ex)
{
actualException = ex;
}
break;
default:
try
{
pairwiseOverlapObj.AlignSimple(aInput, bInput);
}
catch (ArgumentException ex)
{
actualException = ex;
}
break;
}
break;
case AlignParameters.AllParam:
switch (alignType)
{
case AlignmentType.Align:
try
{
pairwiseOverlapObj.Align(sm, gapOpenCost,
gapExtensionCost, aInput, bInput);
}
catch (ArgumentException ex)
{
actualException = ex;
}
break;
default:
try
{
pairwiseOverlapObj.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.Replace("\\r", "").Replace("\\n", "").Replace("\t", ""),
actualException.Message.Replace("\r", "").Replace("\n", "").Replace("\t", ""));
ApplicationLog.WriteLine("PairwiseOverlapAligner P2 : Expected Error message is thrown ", expectedErrorMessage);
}
/// <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>
/// Validates PairwiseOverlapAlignment algorithm for the parameters passed.
/// </summary>
/// <param name="isTextFile">Is text file an input.</param>
/// <param name="alignParam">parameter based on which certain validations are done.</param>
/// <param name="alignType">Is the Align type Simple or Align with Gap Extension cost?</param>
private void ValidatePairwiseOverlapAlignment(bool isTextFile, AlignmentParamType alignParam, AlignmentType alignType)
{
ISequence aInput;
ISequence bInput;
IAlphabet alphabet = Utility.GetAlphabet(this.utilityObj.xmlUtil.GetTextValue(Constants.PairwiseOverlapAlignAlgorithmNodeName, Constants.AlphabetNameNode));
if (isTextFile)
{
// Read the xml file for getting both the files for aligning.
string filePath1 = this.utilityObj.xmlUtil.GetTextValue(Constants.PairwiseOverlapAlignAlgorithmNodeName, Constants.FilePathNode1);
string filePath2 = this.utilityObj.xmlUtil.GetTextValue(Constants.PairwiseOverlapAlignAlgorithmNodeName, Constants.FilePathNode2);
//Parse the files and get the sequence.
var parser = new FastAParser { Alphabet = alphabet };
aInput = parser.Parse(filePath1).ElementAt(0);
bInput = parser.Parse(filePath2).ElementAt(0);
}
else
{
// Read the xml file for getting both the files for aligning.
string origSequence1 = this.utilityObj.xmlUtil.GetTextValue(Constants.PairwiseOverlapAlignAlgorithmNodeName, Constants.SequenceNode1);
string origSequence2 = this.utilityObj.xmlUtil.GetTextValue(Constants.PairwiseOverlapAlignAlgorithmNodeName, Constants.SequenceNode2);
aInput = new Sequence(alphabet, origSequence1);
bInput = new Sequence(alphabet, origSequence2);
}
var aInputString = aInput.ConvertToString();
var bInputString = bInput.ConvertToString();
ApplicationLog.WriteLine(string.Format(null, "PairwiseOverlapAligner BVT : First sequence used is '{0}'.", aInputString));
ApplicationLog.WriteLine(string.Format(null, "PairwiseOverlapAligner BVT : Second sequence used is '{0}'.", bInputString));
string blosumFilePath = this.utilityObj.xmlUtil.GetTextValue(Constants.PairwiseOverlapAlignAlgorithmNodeName, Constants.BlosumFilePathNode);
var sm = new SimilarityMatrix(new StreamReader(blosumFilePath));
int gapOpenCost = int.Parse(this.utilityObj.xmlUtil.GetTextValue(Constants.PairwiseOverlapAlignAlgorithmNodeName, Constants.GapOpenCostNode), null);
int gapExtensionCost = int.Parse(this.utilityObj.xmlUtil.GetTextValue(Constants.PairwiseOverlapAlignAlgorithmNodeName, Constants.GapExtensionCostNode), null);
var pairwiseOverlapObj = new PairwiseOverlapAligner();
if (AlignmentParamType.AllParam != alignParam)
{
pairwiseOverlapObj.SimilarityMatrix = sm;
pairwiseOverlapObj.GapOpenCost = gapOpenCost;
}
IList<IPairwiseSequenceAlignment> result = null;
switch (alignParam)
{
case AlignmentParamType.AlignList:
var sequences = new List<ISequence> {aInput, bInput};
switch (alignType)
{
case AlignmentType.Align:
result = pairwiseOverlapObj.Align(sequences);
break;
default:
result = pairwiseOverlapObj.AlignSimple(sequences);
break;
}
break;
case AlignmentParamType.AlignTwo:
switch (alignType)
{
case AlignmentType.Align:
result = pairwiseOverlapObj.Align(aInput, bInput);
break;
default:
result = pairwiseOverlapObj.AlignSimple(aInput, bInput);
break;
}
break;
case AlignmentParamType.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;
}
// 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(Constants.PairwiseOverlapAlignAlgorithmNodeName, Constants.ExpectedGapExtensionScoreNode);
expectedSequence1 = this.utilityObj.xmlUtil.GetTextValue(Constants.PairwiseOverlapAlignAlgorithmNodeName, Constants.ExpectedGapExtensionSequence1Node);
expectedSequence2 = this.utilityObj.xmlUtil.GetTextValue(Constants.PairwiseOverlapAlignAlgorithmNodeName, Constants.ExpectedGapExtensionSequence2Node);
break;
default:
expectedScore = this.utilityObj.xmlUtil.GetTextValue(Constants.PairwiseOverlapAlignAlgorithmNodeName, Constants.ExpectedScoreNode);
expectedSequence1 = this.utilityObj.xmlUtil.GetTextValue(Constants.PairwiseOverlapAlignAlgorithmNodeName, Constants.ExpectedSequenceNode1);
expectedSequence2 = this.utilityObj.xmlUtil.GetTextValue(Constants.PairwiseOverlapAlignAlgorithmNodeName, 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)
};
align.PairwiseAlignedSequences.Add(alignedSeq);
}
expectedOutput.Add(align);
Assert.IsTrue(AlignmentHelpers.CompareAlignment(result, expectedOutput));
ApplicationLog.WriteLine(string.Format(null, "PairwiseOverlapAligner BVT : Final Score '{0}'.", expectedScore));
ApplicationLog.WriteLine(string.Format(null, "PairwiseOverlapAligner BVT : Aligned First Sequence is '{0}'.", expectedSequence1));
ApplicationLog.WriteLine(string.Format(null, "PairwiseOverlapAligner BVT : Aligned Second Sequence is '{0}'.", expectedSequence2));
}