/// <summary>
/// Gets the design matrix used for obtaining a single abundance value per run.
/// The contrast matrix ends up having the same number of rows as the number of runs.
/// </summary>
public static DesignMatrix GetRunQuantificationDesignMatrix(FoldChangeDataSet dataSet)
{
var columnNames = new List<string>();
var matrixColumns = new List<double[]>();
var contrastValues = new List<double[]>();
var featuresByRun = Enumerable.Range(0, dataSet.RunCount)
.Select(dataSet.GetFeaturesForRun).ToArray();
columnNames.Add("Intercept");
matrixColumns.Add(Enumerable.Repeat(1.0, dataSet.RowCount).ToArray());
contrastValues.Add(Enumerable.Repeat(1.0, dataSet.RunCount).ToArray());
for (int iFeature = 1; iFeature < dataSet.FeatureCount; iFeature++)
{
columnNames.Add("Feature" + iFeature);
matrixColumns.Add(Enumerable.Range(0, dataSet.RowCount).Select(iRow => iFeature == dataSet.Features[iRow] ? 1.0 : 0.0).ToArray());
contrastValues.Add(Enumerable.Range(0, dataSet.RunCount).Select(run =>
{
var features = featuresByRun[run];
if (features.Contains(iFeature))
{
return 1.0/features.Count;
}
return 0;
}).ToArray());
}
for (int iRun = 1; iRun < dataSet.RunCount; iRun++)
{
columnNames.Add("Run" + iRun);
matrixColumns.Add(Enumerable.Range(0, dataSet.RowCount).Select(iRow=>iRun == dataSet.Runs[iRow] ? 1.0 : 0.0).ToArray());
contrastValues.Add(Enumerable.Range(0, dataSet.RunCount).Select(run=>run == iRun ? 1.0 : 0).ToArray());
}
return new DesignMatrix(dataSet, matrixColumns.ToArray(), contrastValues.ToArray(), columnNames.ToArray());
}
private DesignMatrix(FoldChangeDataSet dataSet, double[][] matrixColumns, double[][] contrastValues, string[] columnNames)
{
DataSet = dataSet;
_matrixColumns = matrixColumns;
_contrastValues = contrastValues;
_columnNames = columnNames;
}
/// <summary>
/// Gets the design matrix used for obtaining a single abundance value per run.
/// The contrast matrix ends up having the same number of rows as the number of runs.
/// </summary>
public static DesignMatrix GetRunQuantificationDesignMatrix(FoldChangeDataSet dataSet)
{
var columnNames = new List <string>();
var matrixColumns = new List <double[]>();
var contrastValues = new List <double[]>();
var featuresByRun = Enumerable.Range(0, dataSet.RunCount)
.Select(dataSet.GetFeaturesForRun).ToArray();
columnNames.Add("Intercept");
matrixColumns.Add(Enumerable.Repeat(1.0, dataSet.RowCount).ToArray());
contrastValues.Add(Enumerable.Repeat(1.0, dataSet.RunCount).ToArray());
for (int iFeature = 1; iFeature < dataSet.FeatureCount; iFeature++)
{
columnNames.Add("Feature" + iFeature);
matrixColumns.Add(Enumerable.Range(0, dataSet.RowCount).Select(iRow => iFeature == dataSet.Features[iRow] ? 1.0 : 0.0).ToArray());
contrastValues.Add(Enumerable.Range(0, dataSet.RunCount).Select(run =>
{
var features = featuresByRun[run];
if (features.Contains(iFeature))
{
return(1.0 / features.Count);
}
return(0);
}).ToArray());
}
for (int iRun = 1; iRun < dataSet.RunCount; iRun++)
{
columnNames.Add("Run" + iRun);
matrixColumns.Add(Enumerable.Range(0, dataSet.RowCount).Select(iRow => iRun == dataSet.Runs[iRow] ? 1.0 : 0.0).ToArray());
contrastValues.Add(Enumerable.Range(0, dataSet.RunCount).Select(run => run == iRun ? 1.0 : 0).ToArray());
}
return(new DesignMatrix(dataSet, matrixColumns.ToArray(), contrastValues.ToArray(), columnNames.ToArray()));
}
private DesignMatrix(FoldChangeDataSet dataSet, double[][] matrixColumns, double[][] contrastValues, string[] columnNames)
{
DataSet = dataSet;
_matrixColumns = matrixColumns;
_contrastValues = contrastValues;
_columnNames = columnNames;
}
/// <summary>
/// Constructs a design matrix for a FoldChangeDataSet.
/// If <paramref name="includeFeatureInteraction"/> is true, then the model is:
/// ABUNDANCE ~ FEATURE + SUBJECT + GROUP + FEATURE:GROUP
/// if false, then the model is:
/// ABUNDANCE ~ FEATURE + SUBJECT + GROUP
/// </summary>
/// <param name="dataSet"></param>
/// <param name="includeFeatureInteraction"></param>
/// <returns></returns>
public static DesignMatrix GetDesignMatrix(FoldChangeDataSet dataSet, bool includeFeatureInteraction)
{
var columnNames = new List <string>();
var matrixColumns = new List <double[]>();
var contrastValues = new List <double>();
columnNames.Add("Intercept");
matrixColumns.Add(Enumerable.Repeat(1.0, dataSet.RowCount).ToArray());
contrastValues.Add(0);
for (int iFeature = 1; iFeature < dataSet.FeatureCount; iFeature++)
{
columnNames.Add("Feature" + iFeature);
matrixColumns.Add(Enumerable.Range(0, dataSet.RowCount).Select(iRow => iFeature == dataSet.Features[iRow] ? 1.0 : 0.0).ToArray());
contrastValues.Add(0.0);
}
if (dataSet.RowCount != dataSet.SubjectCount)
{
int controlGroupCount = dataSet.SubjectControls.Count(f => f);
var subjectIndexes = Enumerable.Range(1, dataSet.SubjectCount - 1);
foreach (int iSubject in subjectIndexes)
{
columnNames.Add("Subject" + iSubject);
matrixColumns.Add(Enumerable.Range(0, dataSet.RowCount).Select(iRow => iSubject == dataSet.Subjects[iRow] ? 1.0 : 0.0).ToArray());
if (dataSet.IsSubjectInControlGroup(iSubject))
{
contrastValues.Add(-1.0 / controlGroupCount);
}
else
{
contrastValues.Add(1.0 / (dataSet.SubjectCount - controlGroupCount));
}
}
}
columnNames.Add("Control");
matrixColumns.Add(Enumerable.Range(0, dataSet.RowCount).Select(iRow => dataSet.IsRowInControlGroup(iRow) ? 1.0 : 0.0).ToArray());
contrastValues.Add(-1.0);
if (includeFeatureInteraction)
{
for (int iFeature = 1; iFeature < dataSet.FeatureCount; iFeature++)
{
columnNames.Add("Feature" + iFeature + "Interaction");
matrixColumns.Add(Enumerable.Range(0, dataSet.RowCount).Select(iRow =>
iFeature == dataSet.Features[iRow] && dataSet.IsRowInControlGroup(iRow) ? 1.0 : 0.0).ToArray());
contrastValues.Add(-1.0 / dataSet.FeatureCount);
}
}
return(new DesignMatrix(dataSet, matrixColumns.ToArray(), contrastValues.Select(value => new[] { value }).ToArray(), columnNames.ToArray()));
}
/// <summary>
/// Constructs a design matrix for a FoldChangeDataSet.
/// If <paramref name="includeFeatureInteraction"/> is true, then the model is:
/// ABUNDANCE ~ FEATURE + SUBJECT + GROUP + FEATURE:GROUP
/// if false, then the model is:
/// ABUNDANCE ~ FEATURE + SUBJECT + GROUP
/// </summary>
/// <param name="dataSet"></param>
/// <param name="includeFeatureInteraction"></param>
/// <returns></returns>
public static DesignMatrix GetDesignMatrix(FoldChangeDataSet dataSet, bool includeFeatureInteraction)
{
var columnNames = new List<string>();
var matrixColumns = new List<double[]>();
var contrastValues = new List<double>();
columnNames.Add("Intercept");
matrixColumns.Add(Enumerable.Repeat(1.0, dataSet.RowCount).ToArray());
contrastValues.Add(0);
for (int iFeature = 1; iFeature < dataSet.FeatureCount; iFeature++)
{
columnNames.Add("Feature" + iFeature);
matrixColumns.Add(Enumerable.Range(0, dataSet.RowCount).Select(iRow=>iFeature == dataSet.Features[iRow] ? 1.0 : 0.0).ToArray());
contrastValues.Add(0.0);
}
if (dataSet.RowCount != dataSet.SubjectCount)
{
int controlGroupCount = dataSet.SubjectControls.Count(f => f);
var subjectIndexes = Enumerable.Range(1, dataSet.SubjectCount - 1);
foreach (int iSubject in subjectIndexes)
{
columnNames.Add("Subject" + iSubject);
matrixColumns.Add(Enumerable.Range(0, dataSet.RowCount).Select(iRow=>iSubject == dataSet.Subjects[iRow] ? 1.0 : 0.0).ToArray());
if (dataSet.IsSubjectInControlGroup(iSubject))
{
contrastValues.Add(-1.0/controlGroupCount);
}
else
{
contrastValues.Add(1.0 / (dataSet.SubjectCount - controlGroupCount));
}
}
}
columnNames.Add("Control");
matrixColumns.Add(Enumerable.Range(0, dataSet.RowCount).Select(iRow => dataSet.IsRowInControlGroup(iRow) ? 1.0 : 0.0).ToArray());
contrastValues.Add(-1.0);
if (includeFeatureInteraction)
{
for (int iFeature = 1; iFeature < dataSet.FeatureCount; iFeature++)
{
columnNames.Add("Feature" + iFeature + "Interaction");
matrixColumns.Add(Enumerable.Range(0, dataSet.RowCount).Select(iRow =>
iFeature == dataSet.Features[iRow] && dataSet.IsRowInControlGroup(iRow) ? 1.0 : 0.0).ToArray());
contrastValues.Add(-1.0 / dataSet.FeatureCount);
}
}
return new DesignMatrix(dataSet, matrixColumns.ToArray(), contrastValues.Select(value => new[]{value}).ToArray(), columnNames.ToArray());
}