/// <summary>
/// Generate a symmetric distance matrix from a set of unaligned sequences.
/// </summary>
/// <param name="sequences">a set of unaligned sequences</param>
public void GenerateDistanceMatrix(IList <ISequence> sequences)
{
// Generate k-mer counting dictionary for each sequence
try
{
_allCountsDictionary = new Dictionary <string, float> [sequences.Count];
Parallel.For(0, sequences.Count, i =>
{
Dictionary <string, float> currentDictionary = KmerDistanceScoreCalculator.CalculateKmerCounting(sequences[i], _kmerLength);
MsaUtils.Normalize(currentDictionary);
_allCountsDictionary[i] = currentDictionary;
});
}
catch (OutOfMemoryException ex)
{
throw new Exception("Out of memory when generating kmer counting", ex.InnerException);
}
// Construct a SymmetricDistanceMatrix
// with dimension equals to the number of sequences
_distanceMatrix = new SymmetricDistanceMatrix(sequences.Count);
// Fill in DistanceMatrix
Parallel.For(1, sequences.Count, PAMSAMMultipleSequenceAligner.parallelOption, row =>
{
for (int col = 0; col < row; ++col)
{
float distanceScore = _kmerScoreCalculator.CalculateDistanceScore
(_allCountsDictionary[row], _allCountsDictionary[col]);
_distanceMatrix[row, col] = distanceScore;
_distanceMatrix[col, row] = distanceScore;
}
});
}
/// <summary>
/// Construct DistanceMatrix via k-mer counting algorithm
/// </summary>
/// <param name="sequences">a list of unaligned sequences</param>
/// <param name="kmerLength">positive integer length of kmer</param>
/// <param name="moleculeType">moleculeType: DNA, RNA or Protein</param>
/// <param name="distanceFunctionName">distance function name</param>
public KmerDistanceMatrixGenerator(IList <ISequence> sequences, int kmerLength, MoleculeType moleculeType, DistanceFunctionTypes distanceFunctionName)
{
if (sequences.Count == 0)
{
throw new ArgumentException("Empty input sequence list");
}
_kmerLength = kmerLength;
_kmerScoreCalculator = new KmerDistanceScoreCalculator(kmerLength, moleculeType, distanceFunctionName);
GenerateDistanceMatrix(sequences);
}