/// <summary>
/// Construct an aligner
/// </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)
{
Performance.Start();
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");
}
PAMSAMMultipleSequenceAligner.parallelOption = new ParallelOptions {
MaxDegreeOfParallelism = degreeOfParallelism
};
if (numberOfPartitions <= 0)
{
throw new ArgumentException("Invalid number of partition parameter");
}
_numberOfPartitions = numberOfPartitions;
// Validate data type
_alphabet = sequences[0].Alphabet;
Parallel.For(1, sequences.Count, PAMSAMMultipleSequenceAligner.parallelOption, i =>
{
if (!Alphabets.CheckIsFromSameBase(sequences[i].Alphabet, _alphabet))
{
throw new ArgumentException("Inconsistent sequence alphabet");
}
});
List <String> similarityMatrixDNA = new List <String>();
similarityMatrixDNA.Add("AmbiguousDNA");
List <String> similarityMatrixRNA = new List <String>();
similarityMatrixRNA.Add("AmbiguousRNA");
List <String> similarityMatrixProtein = new List <String>();
similarityMatrixProtein.Add("BLOSUM45");
similarityMatrixProtein.Add("BLOSUM50");
similarityMatrixProtein.Add("BLOSUM62");
similarityMatrixProtein.Add("BLOSUM80");
similarityMatrixProtein.Add("BLOSUM90");
similarityMatrixProtein.Add("PAM250");
similarityMatrixProtein.Add("PAM30");
similarityMatrixProtein.Add("PAM70");
if (_alphabet is DnaAlphabet)
{
if (!similarityMatrixDNA.Contains(similarityMatrix.Name))
{
throw new ArgumentException("Inconsistent similarity matrix");
}
}
else if (_alphabet is ProteinAlphabet)
{
if (!similarityMatrixProtein.Contains(similarityMatrix.Name))
{
throw new ArgumentException("Inconsistent similarity matrix");
}
}
else if (_alphabet is RnaAlphabet)
{
if (!similarityMatrixRNA.Contains(similarityMatrix.Name))
{
throw new ArgumentException("Inconsistent similarity matrix");
}
}
else
{
throw new ArgumentException("Invalid alphabet");
}
// Initialize parameters
_kmerLength = kmerLength;
_distanceFunctionName = distanceFunctionName;
_hierarchicalClusteringMethodName = hierarchicalClusteringMethodName;
_profileAlignerName = profileAlignerMethodName;
_profileProfileFunctionName = profileFunctionName;
SimilarityMatrix = similarityMatrix;
GapOpenCost = gapOpenPenalty;
GapExtensionCost = gapExtendPenalty;
MsaUtils.SetProfileItemSets(_alphabet);
Performance.Snapshot("Start Aligning");
// Work...
Align(sequences);
}
/// <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)
{
Performance.Start();
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");
}
//_degreeOfParallelism = degreeOfParallelism;
parallelOption = new ParallelOptions {
MaxDegreeOfParallelism = degreeOfParallelism
};
if (numberOfPartitions <= 0)
{
throw new ArgumentException("Invalid number of partition parameter");
}
_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(_alphabet);
Performance.Snapshot("Start Aligning");
// Work...
DoAlignment(sequences);
}
