public List <NamedSequence> ConvertToSequences(Matrix <string, string, SufficientStatistics> matrix)
{
bool isMultistate = matrix.RowKeys.Any(key => key.Contains('#'));
if (!isMultistate)
{
matrix = new BinaryToMultistateView <string, string, SufficientStatistics>(matrix, Tabulate.BinaryToMultistateMapping(matrix), MBT.Escience.ValueConverter.SufficientStatisticsToInt);
}
var rowKeysAsMerAndPos = (from key in matrix.RowKeys
let merAndPos = Tabulate.GetMerAndPos(key)
orderby merAndPos.Value
select new
{
Position0 = (int)merAndPos.Value - 1,
Chars = merAndPos.Key.Split('#'),
Key = key
}).ToList();
List <NamedSequence> result = new List <NamedSequence>(matrix.ColCount);
for (int pid = 0; pid < matrix.ColCount; pid++)
{
StringBuilder seq = new StringBuilder(matrix.RowCount);
foreach (var rowKey in rowKeysAsMerAndPos)
{
Helper.CheckCondition(seq.Length <= rowKey.Position0, "There appears to be multiple keys with the same position.");
while (seq.Length < rowKey.Position0)
{
seq.Append(MissingChar);
}
DiscreteStatistics stats = matrix.GetValueOrMissing(rowKey.Key, matrix.ColKeys[pid]).AsDiscreteStatistics();
if (stats.IsMissing())
{
seq.Append(MissingChar);
}
else
{
string thisChar;
if (rowKey.Chars.Length == 1)
{
thisChar = rowKey.Chars[0];
Helper.CheckCondition((int)stats == 1);
}
else
{
thisChar = rowKey.Chars[stats];
}
Helper.CheckCondition(thisChar.Length == 1, "State {0} is too long. Must be a single character.", thisChar);
seq.Append(thisChar);
}
}
result.Add(new NamedSequence(matrix.ColKeys[pid], seq.ToString()));
}
return(result);
}
public void ValidateDiscreteStatisticsGetMissingInstance()
{
DiscreteStatistics discreteStat = DiscreteStatistics.GetMissingInstance;
Assert.IsTrue(discreteStat.IsMissing());
}