private IEnumerable <RowKeyColKeyValue <string, string, UOPair <char> > > TripleEnumerable(HashSet <string> cidExcludeList, HashSet <string> snpExcludeSet)
{
CounterWithMessages counterWithMessages = new CounterWithMessages(SnpFile, compressionRatio: GZ ? .9 : 0);
using (TextReader textReader =
#if !SILVERLIGHT
GZ ? SnpFile.UnGZip() :
#endif
SnpFile.OpenText())
{
string line;
while (null != (line = textReader.ReadLine()))
{
counterWithMessages.Increment();
string[] field = line.Split('\t');
Helper.CheckCondition(field.Length == 4, "Expect lines of snp file to have four fields. " + line);
string snp = field[0];
string cid = field[1];
string value = field[2];
double confidence = double.Parse(field[3]);
Helper.CheckCondition(value.Length == 2 && value.All(c => "ACTG".Contains(c)), () => "Expect values in snp file to be a pair of ACT or G. " + value);
if (cidExcludeList.Contains(cid) || confidence < MissingThreshold || snpExcludeSet.Contains(snp))
{
continue; //not break;
}
yield return(RowKeyColKeyValue.Create(snp, cid, UOPair.Create(value[0], value[1])));
}
}
counterWithMessages.Finished();
}
public Matrix <string, string, double> ToKernel(Matrix <string, string, double> unnormalizedInput, int?cidInBatchCountOrNull = null)
{
if (null == cidInBatchCountOrNull)
{
return(ToKernel(unnormalizedInput));
}
var input = RowNormalizer.Normalize(unnormalizedInput);
var cidListList = SpecialFunctions.DivideListIntoEqualChunksFromChunkSize <string>(input.ColKeys, cidInBatchCountOrNull.Value);
Console.WriteLine("cids divided into {0} batches of about {1}", cidListList.Count, cidInBatchCountOrNull);
Helper.CheckCondition(cidListList.Sum(l => l.Count) == input.ColCount, "real assert");
var counterWithMessages = new CounterWithMessages("kernel combintations ", 1, (cidListList.Count * cidListList.Count + cidListList.Count) / 2);
var kernelPieces2D = new Matrix <string, string, double> [cidListList.Count, cidListList.Count];
for (int i = 0; i < cidListList.Count; ++i)
{
Console.WriteLine("Loading batch {0}, size {1}x{2}", i, cidListList[i].Count, unnormalizedInput.RowCount);
var matrixI = input.SelectColsView(cidListList[i]).ToShoMatrix(verbose: true);
Parallel.For(i, cidListList.Count, ParallelOptionsScope.Current, j =>
{
Console.WriteLine("Loading batch {0}, size {1}x{2}", j, cidListList[j].Count, unnormalizedInput.RowCount);
if (i == j)
{
ShoMatrix kii = JustKernel(matrixI);
kernelPieces2D[i, i] = kii;
}
else
{
var matrixJ = input.SelectColsView(cidListList[j]).ToShoMatrix(verbose: true);
ShoMatrix kij = JustKernel(matrixI, matrixJ);
kernelPieces2D[i, j] = kij;
kernelPieces2D[j, i] = kij.TransposeView().ToShoMatrix();
}
counterWithMessages.Increment();
});
}
counterWithMessages.Finished();
var output = MatrixExtensions.MergeRowsAndColsView(kernelPieces2D);
Helper.CheckCondition(output.RowKeys.SequenceEqual(output.ColKeys) && output.ColKeys.SequenceEqual(unnormalizedInput.ColKeys), "Assert: MergeRows isn't working as expected");
KernelNormalizeInPlace(ref output);
return(output);
}
public virtual Matrix <string, string, LinearTransform> LinearTransformMatrix(Matrix <string, string, double> inputMatrix)
{
var counterWithMessages = new CounterWithMessages("RowNormalizing ", null, inputMatrix.RowCount, quiet: !Verbose);
var linearTransformMatrix = DenseMatrix <string, string, LinearTransform> .CreateDefaultInstance(inputMatrix.RowKeys, new[] { "" }, null);
Parallel.ForEach(inputMatrix.AppendIndex(), ParallelOptionsScope.Current, unnormalizedListAndIndex =>
{
counterWithMessages.Increment();
var linearTransform = CreateLinearTransform(unnormalizedListAndIndex.Item1, inputMatrix.RowKeys[unnormalizedListAndIndex.Item2]);
linearTransformMatrix[unnormalizedListAndIndex.Item2, 0] = linearTransform;
});
counterWithMessages.Finished();
return(linearTransformMatrix);
}
private static Dictionary <string, List <double> > CreateCidToDoubleList(Matrix <string, string, double> matrix)
{
CounterWithMessages counterWithMessages = new CounterWithMessages("loading cid columns ", null, matrix.ColCount, quiet: matrix.ColCount < 10);
var listOfLists = ((IList <IList <double> >)matrix.TransposeView());
var cidToDoubleList =
(from colKey in matrix.ColKeys
.AsParallel().WithParallelOptionsScope()
let colIndex = matrix.IndexOfColKey[colKey]
let doubleList = CreateDoubleList(colIndex, listOfLists, counterWithMessages)
select Tuple.Create(colKey, doubleList)
).ToDictionary();
counterWithMessages.Finished();
return(cidToDoubleList);
}
virtual public Matrix <string, string, T> Filter <T>(Matrix <string, string, T> predictorIn, Matrix <string, string, T> target)
{
var counterWithMessages = new CounterWithMessages("rowFilter " + ToString(), null, predictorIn.RowCount, quiet: !Verbose);
var goodRowKeySet =
(
from rowKey in predictorIn.RowKeys
.AsParallel().WithParallelOptionsScope()
where AlwaysTrue(counterWithMessages) && IsGood(predictorIn.SelectRowsView(rowKey), target)
select rowKey
).ToHashSet();
counterWithMessages.Finished();
var predictorOut = predictorIn.SelectRowsView(predictorIn.RowKeys.Intersect(goodRowKeySet));
return(predictorOut);
}
private IEnumerable <RowKeyColKeyValue <string, string, UOPair <char> > > TripleEnumerable()
{
//int? totalLineCountOrNull = null;
//int? messageIntervalOrNull = 10000;
//using (TextReader textReader = SnpFile.OpenText())
//{
// string line = textReader.ReadLine();
// if (null != line || line.Length == 0)
// {
// totalLineCountOrNull = (int?)(SnpFile.Length / (long)(line.Length + 2 /*line end*/));
// messageIntervalOrNull = null;
// }
//}
CounterWithMessages counterWithMessages = new CounterWithMessages(SnpFile); //"Reading " + SnpFile.Name, messageIntervalOrNull, totalLineCountOrNull);
using (TextReader textReader = SnpFile.OpenText())
{
string line;
while (null != (line = textReader.ReadLine()))
{
counterWithMessages.Increment();
string[] field = line.Split('\t');
Helper.CheckCondition(field.Length == 3, "Expect lines of snp file to have three fields. " + line);
string cid = field[0];
string snp = field[1];
string value = field[2];
if (value == "00")
{
continue; //not break;
}
Helper.CheckCondition(value.Length == 2 && value.All(c => "ACTG".Contains(c)), () => "Expect values in snp file to be a pair of ACTG or 00. " + value);
yield return(RowKeyColKeyValue.Create(snp, cid, UOPair.Create(value[0], value[1])));
}
counterWithMessages.Finished();
}
}
