private static void ProcessColMatch(IDataWithAnnotationColumns mdata, Regex regex, string replacement, int col, bool keepColumns, bool semicolons)
{
string[] values = new string[mdata.RowCount];
for (int row = 0; row < mdata.RowCount; row++)
{
string fullString = mdata.StringColumns[col][row];
string[] inputParts = semicolons ? fullString.Split(';') : new[] { fullString };
values[row] = RegexMatchOrReplace(inputParts[0]);
for (int i = 1; i < inputParts.Length; i++)
{
values[row] += ";" + RegexMatchOrReplace(inputParts[i]);
}
}
if (keepColumns)
{
mdata.AddStringColumn(mdata.StringColumnNames[col], null, values);
}
else
{
mdata.StringColumns[col] = values;
}
string RegexMatchOrReplace(string input)
{
if (string.IsNullOrEmpty(replacement))
{
return(regex.Match(input).Groups[1].ToString());
}
return(regex.Replace(input, replacement));
}
}
public void WriteDataWithAnnotationColumnsTest()
{
// main data
IDataWithAnnotationColumns mdata = PerseusFactory.CreateDataWithAnnotationColumns();
// annotation columns
mdata.AddStringColumn("strcol1", "this is stringcol1", new[] { "1", "2" });
mdata.AddStringColumn("strcol2", "", new[] { "", "hallo" });
mdata.AddNumericColumn("numcol", "", new[] { 1.0, 2.0 });
mdata.AddMultiNumericColumn("multnumcol", "this is multnumcol", new[] { new[] { -2.0, 2.0 }, new double[] {} });
mdata.AddCategoryColumn("catcol", "", new[] { new[] { "cat1", "cat1.1" }, new[] { "cat2", "cat1" } });
string mdataStr;
using (MemoryStream memstream = new MemoryStream())
using (StreamWriter writer = new StreamWriter(memstream)) {
PerseusUtils.WriteDataWithAnnotationColumns(mdata, writer);
writer.Flush();
mdataStr = Encoding.UTF8.GetString(memstream.ToArray());
}
IMatrixData mdata3 = PerseusFactory.CreateMatrixData();
PerseusUtils.ReadMatrix(mdata3, new ProcessInfo(new Settings(), status => { }, progress => { }, 1, i => { }),
() => {
StreamReader tmpStream = new StreamReader(new MemoryStream(Encoding.UTF8.GetBytes(mdataStr)));
return(tmpStream);
}, "matrix1", '\t');
IDataWithAnnotationColumns mdata2 = mdata3;
Assert.AreEqual(2, mdata2.RowCount);
Assert.AreEqual(2, mdata2.StringColumnCount);
Assert.AreEqual(1, mdata2.NumericColumnCount);
Assert.AreEqual(1, mdata2.CategoryColumnCount);
Assert.AreEqual(1, mdata2.MultiNumericColumnCount);
Assert.AreEqual("hallo", mdata2.StringColumns[mdata2.StringColumnNames.FindIndex(col => col.Equals("strcol2"))][1]);
}
private static void AddStringColumns(IDataWithAnnotationColumns mdata1, IDataWithAnnotationColumns mdata2,
Parameters parameters, IList <int[]> indexMap, IDataWithAnnotationColumns result)
{
int[] stringCols = parameters.GetParam <int[]>("Text columns").Value;
string[][] newStringColumns = new string[stringCols.Length][];
string[] newStringColNames = new string[stringCols.Length];
for (int i = 0; i < stringCols.Length; i++)
{
string[] oldCol = mdata2.StringColumns[stringCols[i]];
newStringColNames[i] = mdata2.StringColumnNames[stringCols[i]];
newStringColumns[i] = new string[mdata1.RowCount];
for (int j = 0; j < mdata1.RowCount; j++)
{
int[] inds = indexMap[j];
List <string> values = new List <string>();
foreach (int ind in inds)
{
string v = oldCol[ind];
if (v.Length > 0)
{
values.Add(v);
}
}
newStringColumns[i][j] = values.Count == 0 ? "" : StringUtils.Concat(";", values.ToArray());
}
}
for (int i = 0; i < stringCols.Length; i++)
{
result.AddStringColumn(newStringColNames[i], "", newStringColumns[i]);
}
}
public static void ProcessData(IDataWithAnnotationColumns mdata, IAnnotationProvider annotationProvider, Parameters para, ProcessInfo processInfo)
{
string[] baseIds = GetBaseIds(mdata, annotationProvider, para);
bool success = ProcessDataAddAnnotation(mdata.RowCount, annotationProvider, para, baseIds, processInfo, out string[] name,
out int[] catColInds,
out int[] textColInds, out int[] numColInds, out string[][][] catCols, out string[][] textCols,
out double[][] numCols);
if (!success)
{
return;
}
for (int i = 0; i < catCols.Length; i++)
{
mdata.AddCategoryColumn(name[catColInds[i]], "", catCols[i]);
}
for (int i = 0; i < textCols.Length; i++)
{
mdata.AddStringColumn(name[textColInds[i]], "", textCols[i]);
}
for (int i = 0; i < numCols.Length; i++)
{
mdata.AddNumericColumn(name[numColInds[i]], "", numCols[i]);
}
}
public static void CombineRows(this IDataWithAnnotationColumns mdata, List <int> rowIdxs, Func <double[], double> combineNumeric, Func <string[], string> combineString, Func <string[][], string[]> combineCategory, Func <double[][], double[]> combineMultiNumeric)
{
if (!rowIdxs.Any())
{
return;
}
var resultRow = rowIdxs[0];
for (int i = 0; i < mdata.NumericColumnCount; i++)
{
var column = mdata.NumericColumns[i];
var values = ArrayUtils.SubArray(column, rowIdxs);
column[resultRow] = combineNumeric(values);
}
for (int i = 0; i < mdata.StringColumnCount; i++)
{
var column = mdata.StringColumns[i];
var values = ArrayUtils.SubArray(column, rowIdxs);
column[resultRow] = combineString(values);
}
for (int i = 0; i < mdata.CategoryColumnCount; i++)
{
var column = mdata.GetCategoryColumnAt(i);
var values = ArrayUtils.SubArray(column, rowIdxs);
column[resultRow] = combineCategory(values);
mdata.SetCategoryColumnAt(column, i);
}
for (int i = 0; i < mdata.MultiNumericColumnCount; i++)
{
var column = mdata.MultiNumericColumns[i];
var values = ArrayUtils.SubArray(column, rowIdxs);
column[resultRow] = combineMultiNumeric(values);
}
}
private static void AddCategoricalColumns(IDataWithAnnotationColumns mdata1, IDataWithAnnotationColumns mdata2,
Parameters parameters, IList <int[]> indexMap, IDataWithAnnotationColumns result)
{
int[] catCols = parameters.GetParam <int[]>("Categorical columns").Value;
string[][][] newCatColumns = new string[catCols.Length][][];
string[] newCatColNames = new string[catCols.Length];
for (int i = 0; i < catCols.Length; i++)
{
string[][] oldCol = mdata2.GetCategoryColumnAt(catCols[i]);
newCatColNames[i] = mdata2.CategoryColumnNames[catCols[i]];
newCatColumns[i] = new string[mdata1.RowCount][];
for (int j = 0; j < mdata1.RowCount; j++)
{
int[] inds = indexMap[j];
List <string[]> values = new List <string[]>();
foreach (int ind in inds)
{
string[] v = oldCol[ind];
if (v.Length > 0)
{
values.Add(v);
}
}
newCatColumns[i][j] = values.Count == 0
? new string[0]
: ArrayUtils.UniqueValues(ArrayUtils.Concat(values.ToArray()));
}
}
for (int i = 0; i < catCols.Length; i++)
{
result.AddCategoryColumn(newCatColNames[i], "", newCatColumns[i]);
}
}
private static void ReadMatrixDataInto(IDataWithAnnotationColumns data, string file, ProcessInfo processInfo)
{
var mdata = PerseusFactory.CreateMatrixData();
PerseusUtils.ReadMatrixFromFile(mdata, processInfo, file, '\t');
data.CopyAnnotationColumnsFrom(mdata);
}
private static int[][] GetIndexMap(IDataWithAnnotationColumns mdata1, IDataWithAnnotationColumns mdata2,
Parameters parameters, string separator)
{
ParameterWithSubParams <bool> p = parameters.GetParamWithSubParams <bool>("Use additional column pair");
bool addtlCol = p.Value;
Dictionary <string, List <int> > idToCols2 = addtlCol
? GetIdToColsPair(mdata2, parameters, p.GetSubParameters(), separator)
: GetIdToColsSingle(mdata2, parameters);
string[][] matchCol1 = addtlCol
? GetColumnPair(mdata1, parameters, p.GetSubParameters(), separator)
: GetColumnSplitBySemicolon(mdata1, parameters, "Matching column in matrix 1");
int[][] indexMap = new int[matchCol1.Length][];
for (int i = 0; i < matchCol1.Length; i++)
{
List <int> q = new List <int>();
foreach (string s in matchCol1[i])
{
if (idToCols2.ContainsKey(s))
{
q.AddRange(idToCols2[s]);
}
}
indexMap[i] = ArrayUtils.UniqueValues(q.ToArray());
}
return(indexMap);
}
private static void AddIndicator(IDataWithAnnotationColumns result, IData mdata2, int[][] indexMap)
{
string[][] indicatorCol = new string[indexMap.Length][];
for (int i = 0; i < indexMap.Length; i++)
{
indicatorCol[i] = indexMap[i].Length > 0 ? new[] { "+" } : new string[0];
}
result.AddCategoryColumn(mdata2.Name, "", indicatorCol);
}
private static List <string[][]> GetCategoryColumns(IDataWithAnnotationColumns mdata)
{
List <string[][]> result = new List <string[][]>();
for (int i = 0; i < mdata.CategoryColumnCount; i++)
{
result.Add(mdata.GetCategoryColumnAt(i));
}
return(result);
}
/// <summary>
/// Get the string column with the identifiers used in the mapping.
/// </summary>
/// <param name="mdata"></param>
/// <param name="annotationProvider"></param>
/// <param name="para"></param>
/// <returns></returns>
private static string[] GetBaseIds(IDataWithAnnotationColumns mdata, IAnnotationProvider annotationProvider, Parameters para)
{
ParameterWithSubParams <int> spd = para.GetParamWithSubParams <int>("Source");
int ind = spd.Value;
Parameters param = spd.GetSubParameters();
int baseCol = param.GetParam <int>(annotationProvider.Sources[ind].id + " column").Value;
string[] baseIds = mdata.StringColumns[baseCol];
return(baseIds);
}
private static string[] GetBaseIds(Parameters para, IDataWithAnnotationColumns mdata)
{
PerseusUtils.GetAvailableAnnots(out string[] baseNames, out AnnotType[][] types, out string[] files);
ParameterWithSubParams <int> spd = para.GetParamWithSubParams <int>("Source");
int ind = spd.Value;
Parameters param = spd.GetSubParameters();
int baseCol = param.GetParam <int>(baseNames[ind] + " column").Value;
string[] baseIds = mdata.StringColumns[baseCol];
return(baseIds);
}
private static string[][] GetColumnPair(IDataWithAnnotationColumns mdata1, Parameters parameters, Parameters subPar,
string separator)
{
string[][] matchCol = GetColumnSplitBySemicolon(mdata1, parameters, "Matching column in matrix 1");
string[][] matchColAddtl = GetColumnSplitBySemicolon(mdata1, subPar, "Additional column in matrix 1");
string[][] result = new string[matchCol.Length][];
for (int i = 0; i < result.Length; i++)
{
result[i] = Combine(matchCol[i], matchColAddtl[i], separator);
}
return(result);
}
/// <summary>
/// Creates a default implementation of <see cref="INetworkInfo"/> from the given graph
/// and node/edge tables and indices.
/// </summary>
public static INetworkInfo CreateNetworkInfo(IGraph graph, IDataWithAnnotationColumns nodeTable,
Dictionary <INode, int> nodeIndex,
IDataWithAnnotationColumns edgeTable, Dictionary <IEdge, int> edgeIndex, string name, Guid guid)
{
var networkInfoTypeName = Assembly.CreateQualifiedName("PerseusLibS", "PerseusLibS.Data.Network.NetworkInfo");
var type = Type.GetType(networkInfoTypeName);
if (type == null)
{
throw new Exception($"Cannot load type {networkInfoTypeName}.");
}
return((INetworkInfo)Activator.CreateInstance(type, graph, nodeTable, nodeIndex, edgeTable, edgeIndex, name, guid));
}
public void UniqueValuesTest()
{
Mock <IDataWithAnnotationColumns> moq = new Moq.Mock <IDataWithAnnotationColumns>();
List <string[]> testList = new List <string[]> {
new[] { "a;b", "a;a" }
};
moq.Setup(data => data.StringColumns).Returns(testList);
IDataWithAnnotationColumns asdf = moq.Object;
asdf.UniqueValues(new[] { 0 });
CollectionAssert.AreEqual(new [] { "a;b", "a" }, testList[0]);
}
private static string[][] GetColumnSplitBySemicolon(IDataWithAnnotationColumns mdata, Parameters parameters,
string colName)
{
string[] matchingColumn2 = mdata.StringColumns[parameters.GetParam <int>(colName).Value];
string[][] w = new string[matchingColumn2.Length][];
for (int i = 0; i < matchingColumn2.Length; i++)
{
string r = matchingColumn2[i].Trim();
w[i] = r.Length == 0 ? new string[0] : r.Split(';');
w[i] = ArrayUtils.UniqueValues(w[i]);
}
return(w);
}
private static void AddMainColumns(IDataWithAnnotationColumns mdata1, IMatrixData mdata2, Parameters parameters,
IList <int[]> indexMap, IMatrixData result)
{
Func <double[], double> avExpression = GetAveraging(parameters.GetParam <int>("Combine main values").Value);
int[] exColInds = parameters.GetParam <int[]>("Main columns").Value;
if (exColInds.Length > 0)
{
double[,] newExColumns = new double[mdata1.RowCount, exColInds.Length];
double[,] newQuality = new double[mdata1.RowCount, exColInds.Length];
bool[,] newIsImputed = new bool[mdata1.RowCount, exColInds.Length];
string[] newExColNames = new string[exColInds.Length];
for (int i = 0; i < exColInds.Length; i++)
{
newExColNames[i] = mdata2.ColumnNames[exColInds[i]];
for (int j = 0; j < mdata1.RowCount; j++)
{
int[] inds = indexMap[j];
List <double> values = new List <double>();
List <double> qual = new List <double>();
List <bool> imp = new List <bool>();
foreach (int ind in inds)
{
double v = mdata2.Values.Get(ind, exColInds[i]);
if (!double.IsNaN(v) && !double.IsInfinity(v))
{
values.Add(v);
if (mdata2.Quality.IsInitialized())
{
double qx = mdata2.Quality.Get(ind, exColInds[i]);
if (!double.IsNaN(qx) && !double.IsInfinity(qx))
{
qual.Add(qx);
}
}
if (mdata2.IsImputed != null)
{
bool isi = mdata2.IsImputed[ind, exColInds[i]];
imp.Add(isi);
}
}
}
newExColumns[j, i] = values.Count == 0 ? double.NaN : avExpression(values.ToArray());
newQuality[j, i] = qual.Count == 0 ? double.NaN : avExpression(qual.ToArray());
newIsImputed[j, i] = imp.Count != 0 && AvImp(imp.ToArray());
}
}
MakeNewNames(newExColNames, result.ColumnNames);
AddMainColumns(result, newExColNames, newExColumns, newQuality, newIsImputed);
}
}
private static HashSet <string> GetAllIds(IDataWithAnnotationColumns mdata, int baseCol)
{
string[] x = mdata.StringColumns[baseCol];
HashSet <string> result = new HashSet <string>();
foreach (string y in x)
{
string[] z = y.Length > 0 ? y.Split(';') : new string[0];
foreach (string q in z)
{
result.Add(q.ToLower());
}
}
return(result);
}
public static void UniqueValues(this IDataWithAnnotationColumns mdata, int[] stringCols)
{
foreach (string[] col in stringCols.Select(stringCol => mdata.StringColumns[stringCol]))
{
for (int i = 0; i < col.Length; i++)
{
string q = col[i];
if (q.Length == 0)
{
continue;
}
string[] w = q.Split(';');
w = ArrayUtils.UniqueValues(w);
col[i] = StringUtils.Concat(";", w);
}
}
}
private static void ProcessCol(IDataWithAnnotationColumns mdata, Regex regex, int col, bool keepColumns,
bool semicolons)
{
string[] values = new string[mdata.RowCount];
for (int row = 0; row < mdata.RowCount; row++){
string fullString = mdata.StringColumns[col][row];
string[] inputParts = semicolons ? fullString.Split(';') : new[]{fullString};
values[row] = regex.Match(inputParts[0]).Groups[1].ToString();
for (int i = 1; i < inputParts.Length; i++){
values[row] += ";" + regex.Match(inputParts[i]).Groups[1];
}
}
if (keepColumns){
mdata.AddStringColumn(mdata.StringColumnNames[col], null, values);
} else{
mdata.StringColumns[col] = values;
}
}
/// <summary>
/// Add a number of empty rows to the table
/// </summary>
public static void AddEmptyRows(this IDataWithAnnotationColumns mdata, int length)
{
for (int i = 0; i < mdata.StringColumnCount; i++)
{
mdata.StringColumns[i] = mdata.StringColumns[i].Concat(Enumerable.Repeat(String.Empty, length)).ToArray();
}
for (int i = 0; i < mdata.NumericColumnCount; i++)
{
mdata.NumericColumns[i] = mdata.NumericColumns[i].Concat(Enumerable.Repeat(Double.NaN, length)).ToArray();
}
for (int i = 0; i < mdata.MultiNumericColumnCount; i++)
{
mdata.MultiNumericColumns[i] = mdata.MultiNumericColumns[i].Concat(Enumerable.Range(0, length).Select(_ => new double[0])).ToArray();
}
for (int i = 0; i < mdata.CategoryColumnCount; i++)
{
mdata.SetCategoryColumnAt(mdata.GetCategoryColumnAt(i).Concat(Enumerable.Repeat(new string[0], length)).ToArray(), i);
}
}
private static Dictionary <string, List <int> > GetIdToColsSingle(IDataWithAnnotationColumns mdata2,
Parameters parameters)
{
string[][] matchCol2 = GetColumnSplitBySemicolon(mdata2, parameters, "Matching column in matrix 2");
Dictionary <string, List <int> > idToCols2 = new Dictionary <string, List <int> >();
for (int i = 0; i < matchCol2.Length; i++)
{
foreach (string s in matchCol2[i])
{
if (!idToCols2.ContainsKey(s))
{
idToCols2.Add(s, new List <int>());
}
idToCols2[s].Add(i);
}
}
return(idToCols2);
}
private static void ProcessCol(IDataWithAnnotationColumns mdata, Regex regex, int col, bool keepColumns,
bool semicolons)
{
string[] values = new string[mdata.RowCount];
for (int row = 0; row < mdata.RowCount; row++)
{
string fullString = mdata.StringColumns[col][row];
string[] inputParts = semicolons ? fullString.Split(';') : new[] { fullString };
values[row] = regex.Match(inputParts[0]).Groups[1].ToString();
for (int i = 1; i < inputParts.Length; i++)
{
values[row] += ";" + regex.Match(inputParts[i]).Groups[1];
}
}
if (keepColumns)
{
mdata.AddStringColumn(mdata.StringColumnNames[col], null, values);
}
else
{
mdata.StringColumns[col] = values;
}
}
private static Dictionary <string, List <int> > GetIdToColsPair(IDataWithAnnotationColumns mdata2, Parameters parameters,
Parameters subPar, string separator)
{
string[][] matchCol = GetColumnSplitBySemicolon(mdata2, parameters, "Matching column in matrix 2");
string[][] matchColAddtl = GetColumnSplitBySemicolon(mdata2, subPar, "Additional column in matrix 2");
Dictionary <string, List <int> > idToCols2 = new Dictionary <string, List <int> >();
for (int i = 0; i < matchCol.Length; i++)
{
foreach (string s1 in matchCol[i])
{
foreach (string s2 in matchColAddtl[i])
{
string id = s1 + separator + s2;
if (!idToCols2.ContainsKey(id))
{
idToCols2.Add(id, new List <int>());
}
idToCols2[id].Add(i);
}
}
}
return(idToCols2);
}
private static void AddNumericColumns(IDataWithAnnotationColumns mdata1, IDataWithAnnotationColumns mdata2,
Parameters parameters, IList <int[]> indexMap, IDataWithAnnotationColumns result)
{
Func <double[], double> avNumerical = GetAveraging(parameters.GetParam <int>("Combine numerical values").Value);
int[] numCols = parameters.GetParam <int[]>("Numerical columns").Value;
if (avNumerical != null)
{
double[][] newNumericalColumns = new double[numCols.Length][];
string[] newNumColNames = new string[numCols.Length];
for (int i = 0; i < numCols.Length; i++)
{
double[] oldCol = mdata2.NumericColumns[numCols[i]];
newNumColNames[i] = mdata2.NumericColumnNames[numCols[i]];
newNumericalColumns[i] = new double[mdata1.RowCount];
for (int j = 0; j < mdata1.RowCount; j++)
{
int[] inds = indexMap[j];
List <double> values = new List <double>();
foreach (int ind in inds)
{
double v = oldCol[ind];
if (!double.IsNaN(v))
{
values.Add(v);
}
}
newNumericalColumns[i][j] = values.Count == 0 ? double.NaN : avNumerical(values.ToArray());
}
}
for (int i = 0; i < numCols.Length; i++)
{
result.AddNumericColumn(newNumColNames[i], "", newNumericalColumns[i]);
}
}
else
{
double[][][] newMultiNumericalColumns = new double[numCols.Length][][];
string[] newMultiNumColNames = new string[numCols.Length];
for (int i = 0; i < numCols.Length; i++)
{
double[] oldCol = mdata2.NumericColumns[numCols[i]];
newMultiNumColNames[i] = mdata2.NumericColumnNames[numCols[i]];
newMultiNumericalColumns[i] = new double[mdata1.RowCount][];
for (int j = 0; j < mdata1.RowCount; j++)
{
int[] inds = indexMap[j];
List <double> values = new List <double>();
foreach (int ind in inds)
{
double v = oldCol[ind];
if (!double.IsNaN(v))
{
values.Add(v);
}
}
newMultiNumericalColumns[i][j] = values.ToArray();
}
}
for (int i = 0; i < numCols.Length; i++)
{
result.AddMultiNumericColumn(newMultiNumColNames[i], "", newMultiNumericalColumns[i]);
}
}
}
public RowNameInfo(IDataWithAnnotationColumns mdata)
{
this.mdata = mdata;
}
private static List<string[][]> GetCategoryColumns(IDataWithAnnotationColumns mdata)
{
List<string[][]> result = new List<string[][]>();
for (int i = 0; i < mdata.CategoryColumnCount; i++){
result.Add(mdata.GetCategoryColumnAt(i));
}
return result;
}
/// <summary>
/// Unsafe shortcut for obtaining a string column.
/// </summary>
private static string[] GetStringColumn(this IDataWithAnnotationColumns data, string colname)
{
return(data.StringColumns[data.StringColumnNames.FindIndex(col => col.ToLower().Equals(colname.ToLower()))]);
}
public static void AddAnnotationColumns(IDataWithAnnotationColumns result, IDataWithAnnotationColumns mdata2,
int[][] indexMap, int[] copyTextColumns, (int[] copy, int combine) numeric, int[] copyCatColumns)
string[] GetColumn(IDataWithAnnotationColumns data, int index) => index < data.StringColumnCount
? data.StringColumns[index]
: data.NumericColumns[index - data.StringColumnCount].Select(Convert.ToString).ToArray();
