public void SetUp()
{
histo = new HistogramChar();
}
public void TearDown()
{
this.histo = null;
}
/// <summary>
/// Recognizes the alphabet of the given character sequence. The method uses
/// a simple heuristics which might fail especially when the sequence is
/// short. Please note, that a DNA or RNA sequence can contain more than
/// the symbols of the four bases.
/// </summary>
/// <param name="characters"> Character sequence. </param>
/// <returns> Returns an alphabet. </returns>
static public Alphabet Recognize(IEnumerable<char> characters)
{
HistogramChar histo = new HistogramChar();
histo.Inc(characters);
if ((
histo.Get('F') + histo.Get('E') + histo.Get('P') + histo.Get('Q') +
histo.Get('I') + histo.Get('L') + histo.Get('F') + histo.Get('E') +
histo.Get('p') + histo.Get('q') + histo.Get('i') + histo.Get('l')) > 0)
return (AlphabetAA.Instance());
if (histo.GetRelative('U') > 0.1 || histo.GetRelative('u') > 0.1)
return (AlphabetRNA.Instance());
if (histo.GetRelative('a') > 0.1 || histo.GetRelative('A') > 0.1)
return (AlphabetDNA.Instance());
return (AlphabetAA.Instance());
}
