/// <summary>
/// Samples the data from the model.
/// </summary>
/// <param name="sequenceCount">The number of sequences to sample.</param>
/// <param name="sequenceLength">The length of a sequence.</param>
/// <param name="motifPresenceProbability">The probability that a sequence will contain the motif.</param>
/// <param name="motif">The position frequency matrix defining the motif.</param>
/// <param name="backgroundDist">The background nucleobase distribution.</param>
/// <param name="sequenceData">The sampled sequences.</param>
/// <param name="motifPositionData">
/// The motif positions in the sampled sequences.
/// If the sequence doesn't contain the motif, the position is set to -1.
/// </param>
private static void SampleMotifData(
int sequenceCount,
int sequenceLength,
double motifPresenceProbability,
DiscreteChar[] motif,
DiscreteChar backgroundDist,
out string[] sequenceData,
out int[] motifPositionData)
{
sequenceData = new string[sequenceCount];
motifPositionData = new int[sequenceCount];
for (int i = 0; i < sequenceCount; ++i)
{
if (Rand.Double() < motifPresenceProbability)
{
motifPositionData[i] = Rand.Int(sequenceLength - motif.Length + 1);
var backgroundBeforeChars = Util.ArrayInit(motifPositionData[i], j => backgroundDist.Sample());
var backgroundAfterChars = Util.ArrayInit(sequenceLength - motif.Length - motifPositionData[i], j => backgroundDist.Sample());
var sampledMotifChars = Util.ArrayInit(motif.Length, j => motif[j].Sample());
sequenceData[i] = new string(backgroundBeforeChars) + new string(sampledMotifChars) + new string(backgroundAfterChars);
}
else
{
motifPositionData[i] = -1;
var background = Util.ArrayInit(sequenceLength, j => backgroundDist.Sample());
sequenceData[i] = new string(background);
}
}
}