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 static Matrix <string, string, SufficientStatistics> ConvertToMatrix(List <NamedSequence> sequences, MixtureSemantics mix, BinaryOrMultistate dataType, bool keepOneValueVariables)
{
var colNames = sequences.Select(s => s.Name);
//var rowNames = (from posAndAa in sequences.SelectMany(s => s.AASeq)
// where !AASeq.IsMissing(posAndAa.Value)
// let pos = posAndAa.Key
// let aas = posAndAa.Value
// from c in aas
// let merAndPos = pos + "@" + c
// orderby pos, c
// select merAndPos).Distinct();
var rowNames = (from posAndAa in sequences.SelectMany(s => s.AASeq)
where posAndAa.Value.Count == 1 && !AASeq.IsMissing(posAndAa.Value)
let pos = posAndAa.Key
let aas = posAndAa.Value
let c = aas.First()
let merAndPos = pos + "@" + c
orderby pos, c
select merAndPos).Distinct();
var posToRowNames = (from row in rowNames
let pos = (int)Tabulate.GetMerAndPos(row).Value
group row by pos into g
select new KeyValuePair <int, List <string> >(g.Key, g.ToList())).ToDictionary();
Matrix <string, string, SufficientStatistics> m = DenseMatrix <string, string, SufficientStatistics> .CreateDefaultInstance(rowNames, colNames, MissingStatistics.GetInstance());
foreach (var seq in sequences)
{
foreach (var posAndAa in seq.AASeq)
{
int pos = posAndAa.Key;
if (!posToRowNames.ContainsKey(pos))
{
Helper.CheckCondition(AASeq.IsMissing(posAndAa.Value), "Something's wrong. We thinking everyone is missing at position {0}, but {1} has {2}", pos, seq.Name, posAndAa.Value);
continue;
}
var relevantRows = posToRowNames[pos];
bool isMissing = AASeq.IsMissing(posAndAa.Value);
var myRows = posAndAa.Value.Select(c => pos + "@" + c).ToList();
foreach (var row in relevantRows)
{
SufficientStatistics value;
if (isMissing)
{
value = MissingStatistics.GetInstance();
}
else if (!myRows.Contains(row))
{
value = (BooleanStatistics)false; //in all cases, this is false
}
else if (myRows.Count == 1)
{
value = (BooleanStatistics)true;
}
else
{
switch (mix)
{
case MixtureSemantics.Any:
value = (BooleanStatistics)true; break; //Any means we say you have both
case MixtureSemantics.Pure:
value = (BooleanStatistics)false; break; //Pure means we say you have neither
case MixtureSemantics.Uncertainty:
value = MissingStatistics.GetInstance(); break; //Uncertainty says we don't know which you have.
case MixtureSemantics.Distribution:
double pTrue = 1.0 / myRows.Count;
value = MultinomialStatistics.GetInstance(new double[] { 1 - pTrue, pTrue });
break;
default:
throw new NotImplementedException("Missing a case.");
}
}
m.SetValueOrMissing(row, seq.Name, value);
}
}
}
if (!keepOneValueVariables)
{
m = m.SelectRowsView(m.RowKeys.Where(row => m.RowView(row).Values.Distinct().Count() > 1));
}
switch (dataType)
{
case BinaryOrMultistate.Binary:
return(m);
case BinaryOrMultistate.Multistate:
return(new BinaryToMultistateView <string, string, SufficientStatistics>(m, Tabulate.BinaryToMultistateMapping(m), ValueConverter.SufficientStatisticsToMultinomial));
default:
throw new NotImplementedException("Missing a case");
}
}
