public void TestConvertNumericToMultiNumeric()
{
var mBase = PerseusFactory.CreateMatrixData();
mBase.AddStringColumn("Id", "", new [] { "n1;n2", "n3" });
var mdata = PerseusFactory.CreateMatrixData(new[, ] {
{ 0.0 }, { 1.0 }, { 2.0 }
});
mdata.AddStringColumn("Id", "", new [] { "n1", "n2", "n3" });
var match = new MatchingRowsByName();
var errString = string.Empty;
var param = match.GetParameters(new [] { mBase, mdata }, ref errString);
param.GetParam <int[]>("Copy main columns").Value = new[] { 0 };
param.GetParam <int>("Combine copied main values").Value = 5;
IMatrixData[] supplTables = null;
IDocumentData[] documents = null;
var result = match.ProcessData(new[] { mBase, mdata }, param, ref supplTables, ref documents,
BaseTest.CreateProcessInfo());
Assert.AreEqual(result.MultiNumericColumnCount, 1);
CollectionAssert.AreEqual(new [] { 0.0, 1.0 }, result.MultiNumericColumns[0][0]);
CollectionAssert.AreEqual(new [] { 2.0 }, result.MultiNumericColumns[0][1]);
}
public void TestSmallExample()
{
var m1 = PerseusFactory.CreateMatrixData(new double[, ] {
{ 0, 1 }, { 2, 3 }
}, new List <string> {
"col 1", "col 2"
});
var m2 = PerseusFactory.CreateMatrixData(new double[, ] {
{ 4, 5 }, { 6, 7 }
}, new List <string> {
"col 2", "col 3"
});
var m = new[] { m1, m2 };
var matching = new MatchingColumnsByName();
var errString = string.Empty;
var parameters = matching.GetParameters(m, ref errString);
Assert.IsTrue(string.IsNullOrEmpty(errString));
IMatrixData[] supplTables = null;
IDocumentData[] documents = null;
var pInfo = BaseTest.CreateProcessInfo();
var result = matching.ProcessData(m, parameters, ref supplTables, ref documents, pInfo);
Assert.IsTrue(string.IsNullOrEmpty(pInfo.ErrString));
Assert.AreEqual(4, result.RowCount);
CollectionAssert.AreEqual(new [] { "col 2", "col 1", "col 3" }, result.ColumnNames);
CollectionAssert.AreEqual(new [] { 1, 3, 4, 6.0 }, result.Values.GetColumn(0).ToArray());
CollectionAssert.AreEqual(new [] { 0, 2, double.NaN, double.NaN }, result.Values.GetColumn(1).ToArray());
CollectionAssert.AreEqual(new [] { double.NaN, double.NaN, 5, 7 }, result.Values.GetColumn(2).ToArray());
}
public void TestSmallExampleWithAnnotationColumns()
{
var m1 = PerseusFactory.CreateMatrixData(new double[, ] {
{ 0, 1 }, { 2, 3 }
}, new List <string> {
"col 1", "col 2"
});
m1.AddStringColumn("m1", "", new[] { "a", "b" });
m1.AddStringColumn("common string column", "", new[] { "c", "d" });
m1.AddNumericColumn("m1", "", new[] { 0, 1.0 });
m1.AddNumericColumn("common numeric column", "", new[] { 2, 3.0 });
m1.AddCategoryColumn("common category column", "", new [] { new [] { "cat1" }, new [] { "cat2", "cat3" } });
var m2 = PerseusFactory.CreateMatrixData(new double[, ] {
{ 4, 5 }, { 6, 7 }
}, new List <string> {
"col 2", "col 3"
});
m2.AddStringColumn("common string column", "", new [] { "e", "f" });
m2.AddStringColumn("m2", "", new [] { "g", "h" });
m2.AddNumericColumn("common numeric column", "", new[] { 4, 5.0 });
m2.AddCategoryColumn("common category column", "", new [] { new [] { "cat2" }, new [] { "cat1", "cat4" } });
var m = new[] { m1, m2 };
var matching = new MatchingColumnsByName();
var errString = string.Empty;
var parameters = matching.GetParameters(m, ref errString);
Assert.IsTrue(string.IsNullOrEmpty(errString));
IMatrixData[] supplTables = null;
IDocumentData[] documents = null;
var pInfo = BaseTest.CreateProcessInfo();
var result = matching.ProcessData(m, parameters, ref supplTables, ref documents, pInfo);
Assert.IsTrue(string.IsNullOrEmpty(pInfo.ErrString));
CollectionAssert.AreEqual(new [] { "common string column", "m1", "m2" }, result.StringColumnNames);
CollectionAssert.AreEqual(new [] { "c", "d", "e", "f" }, result.StringColumns[0]);
CollectionAssert.AreEqual(new [] { "a", "b", "", "" }, result.StringColumns[1]);
CollectionAssert.AreEqual(new [] { "", "", "g", "h" }, result.StringColumns[2]);
CollectionAssert.AreEqual(new [] { "common numeric column", "m1" }, result.NumericColumnNames);
CollectionAssert.AreEqual(new [] { 2, 3, 4, 5.0 }, result.NumericColumns[0]);
CollectionAssert.AreEqual(new [] { 0, 1, double.NaN, double.NaN }, result.NumericColumns[1]);
CollectionAssert.AreEqual(new [] { "common category column" }, result.CategoryColumnNames);
var actual = result.GetCategoryColumnAt(0);
var expected = new[] { new[] { "cat1" }, new[] { "cat2", "cat3" }, new[] { "cat2" }, new[] { "cat1", "cat4" } };
Assert.AreEqual(expected.Length, actual.Length);
for (int i = 0; i < actual.Length; i++)
{
CollectionAssert.AreEqual(expected[i], actual[i]);
}
}
public void TestReadEmptyMatrixFromFile()
{
var data = PerseusFactory.CreateDataWithAnnotationColumns();
PerseusUtils.ReadDataWithAnnotationColumns(data, BaseTest.CreateProcessInfo(), () =>
{
var memstream = new MemoryStream(Encoding.UTF8.GetBytes("Node\n#!{Type}T\n"));
return(new StreamReader(memstream));
}, "test", '\t');
Assert.AreEqual(0, data.RowCount);
}
public void TestInitialize()
{
double[,] peptidesValues = { { 9.0 } };
peptides = PerseusFactory.CreateMatrixData(peptidesValues, new List <string> {
"pep_MS/MS Count"
});
peptides.AddNumericColumn("pep_Intensity", "", new [] { 0.0 });
peptides.AddStringColumn("pep_id", "", new [] { "35" });
peptides.AddStringColumn("pep_Protein group IDs", "", new [] { "13;21" });
peptides.Quality.Init(1, 1);
peptides.Quality.Set(0, 0, 1);
ExpandMultiNumeric multiNum = new ExpandMultiNumeric();
string errorString = string.Empty;
Parameters parameters2 = multiNum.GetParameters(peptides, ref errorString);
parameters2.GetParam <int[]>("Text columns").Value = new[] { 1 };
IMatrixData[] suppl = null;
IDocumentData[] docs = null;
multiNum.ProcessData(peptides, parameters2, ref suppl, ref docs, BaseTest.CreateProcessInfo());
double[,] proteinMainValues =
{
{ 166250000.0 },
{ 8346000.0 }
};
proteinMain = PerseusFactory.CreateMatrixData(proteinMainValues, new List <string> {
"prot_LFQ intensity"
});
proteinMain.Name = "protein main";
proteinMain.AddStringColumn("prot_id", "", new [] { "13", "21" });
proteinMain.AddStringColumn("prot_gene name", "", new [] { "geneA", "geneB" });
double[,] expandValues =
{
{ 9.0 },
{ 9.0 }
};
expand = PerseusFactory.CreateMatrixData(expandValues, new List <string> {
"pep_MS/MS Count"
});
expand.Name = "expand";
expand.AddNumericColumn("pep_Intensity", "", new [] { 0.0, 0.0 });
expand.AddStringColumn("pep_id", "", new [] { "35", "35" });
expand.AddStringColumn("pep_Protein group IDs", "", new [] { "13", "21" });
matching = new MatchingRowsByName();
string err = string.Empty;
parameters = matching.GetParameters(new[] { expand, proteinMain }, ref err);
}
public void TestSmallExample()
{
double[,] values = new[, ]
{
{ 0.0, 1.0, 0, 5 },
{ 2.0, 3.0, 0, 5 }
};
IMatrixData mdata = PerseusFactory.CreateMatrixData(values, new List <string> {
"Col___1", "Col___2", "Col___3", "No expand"
});
mdata.ColumnDescriptions = new List <string> {
"Description Col", "Col", "Col", "Description No expand"
};
double[][] multiNum = new[]
{
new[] { 0.0, 1.0 },
new[] { 2.0 }
};
mdata.AddMultiNumericColumn("MultiNum", "", multiNum);
string[] stringCol = new[] { "row1", "row2" };
mdata.AddStringColumn("String", "", stringCol);
ExpandSiteTable expand = new ExpandSiteTable();
IMatrixData[] supplData = null;
IDocumentData[] docs = null;
expand.ProcessData(mdata, new Parameters(), ref supplData, ref docs, BaseTest.CreateProcessInfo());
Assert.AreEqual(2, mdata.ColumnCount);
CollectionAssert.AreEqual(new [] { "No expand", "Col" }, mdata.ColumnNames.ToArray());
Assert.AreEqual(2, mdata.ColumnDescriptions.Count);
CollectionAssert.AreEqual(new [] { "Description No expand", "Description Col" }, mdata.ColumnDescriptions.ToArray());
Assert.AreEqual(6, mdata.RowCount);
Assert.AreEqual(2, mdata.StringColumnCount);
CollectionAssert.AreEqual(new [] { "String", "Unique identifier" }, mdata.StringColumnNames);
CollectionAssert.AreEqual(stringCol.Concat(stringCol).Concat(stringCol).ToArray(), mdata.StringColumns[0]);
Assert.AreEqual(1, mdata.MultiNumericColumnCount);
CollectionAssert.AreEqual(multiNum.Concat(multiNum).Concat(multiNum).ToArray(), mdata.MultiNumericColumns[0]);
}
public void TestSmallExample()
{
SingleChoiceParam matchColParam1 = (SingleChoiceParam)parameters.GetParam <int>("Matching column in table 1");
CollectionAssert.AreEqual(new [] { "pep_id", "pep_Protein group IDs", "pep_Intensity" }, matchColParam1.Values.ToArray());
matchColParam1.Value = 1;
Assert.AreEqual("pep_Protein group IDs", matchColParam1.StringValue);
IMatrixData[] supplTables = null;
IDocumentData[] documents = null;
IMatrixData matched = matching.ProcessData(new[] { expand, proteinMain }, parameters, ref supplTables, ref documents, BaseTest.CreateProcessInfo());
CollectionAssert.AreEqual(new [] { "pep_MS/MS Count", "pep_id", "pep_Protein group IDs", "pep_Intensity" },
matched.ColumnNames.Concat(matched.StringColumnNames).Concat(matched.NumericColumnNames).ToArray());
Assert.AreEqual(2, matched.RowCount);
Assert.AreEqual(1, matched.ColumnCount);
Assert.AreEqual(1, matched.NumericColumnCount);
}
public void TestMatchingCaseInSensitive()
{
var mBase = PerseusFactory.CreateMatrixData();
mBase.AddStringColumn("Name", "", new [] { "A", "a", "B", "b", "C", "c" });
Assert.IsTrue(mBase.IsConsistent(out var mBaseConsistent), mBaseConsistent);
var mdata = PerseusFactory.CreateMatrixData();
mdata.AddStringColumn("Name", "", new [] { "a", "B" });
Assert.IsTrue(mdata.IsConsistent(out var mdataConsistent), mdataConsistent);
var match = new MatchingRowsByName();
var errString = string.Empty;
var param = match.GetParameters(new [] { mBase, mdata }, ref errString);
param.GetParam <bool>("Add indicator").Value = true;
param.GetParam <bool>("Ignore case").Value = true;
IMatrixData[] supplTables = null;
IDocumentData[] documents = null;
var result = match.ProcessData(new[] { mBase, mdata }, param, ref supplTables, ref documents, BaseTest.CreateProcessInfo());
var indicator = result.GetCategoryColumnAt(0).Select(cats => cats.SingleOrDefault() ?? "").ToArray();
CollectionAssert.AreEqual(new [] { "+", "+", "+", "+", "", "" }, indicator);
}
public void TestConvertNumericToMultiNumeric()
{
var mBase = PerseusFactory.CreateMatrixData();
mBase.AddStringColumn("Id", "", new [] { "n1;n2", "n3", "n5" });
Assert.IsTrue(mBase.IsConsistent(out var mBaseConsistent), mBaseConsistent);
var mdata = PerseusFactory.CreateMatrixData(new[, ] {
{ 0.0 }, { 1.0 }, { 2.0 }, { 3.0 }
});
mdata.AddStringColumn("Id", "", new [] { "n1", "n2", "n3", "n4" });
Assert.IsTrue(mdata.IsConsistent(out var mdataConsistent), mdataConsistent);
var match = new MatchingRowsByName();
var errString = string.Empty;
var param = match.GetParameters(new [] { mBase, mdata }, ref errString);
param.GetParam <int[]>("Copy main columns").Value = new[] { 0 };
param.GetParam <int>("Combine copied main values").Value = 5;
param.GetParam <int>("Join style").Value = 1;
param.GetParam <bool>("Add indicator").Value = true;
param.GetParam <bool>("Add original row numbers").Value = true;
IMatrixData[] supplTables = null;
IDocumentData[] documents = null;
var result = match.ProcessData(new[] { mBase, mdata }, param, ref supplTables, ref documents, BaseTest.CreateProcessInfo());
var indicator = result.GetCategoryColumnAt(0).Select(cats => cats.SingleOrDefault() ?? "");
CollectionAssert.AreEqual(new [] { "+", "+", "", "+" }, indicator);
CollectionAssert.AreEqual(new [] { "n1;n2", "n3", "n5", "n4" }, result.GetStringColumn("Id"));
CollectionAssert.AreEqual(new [] { "Original row numbers", "Column 1" }, result.MultiNumericColumnNames);
CollectionAssert.AreEqual(new [] { 0.0, 1.0 }, result.MultiNumericColumns[0][0]);
CollectionAssert.AreEqual(new [] { 2.0 }, result.MultiNumericColumns[0][1]);
CollectionAssert.AreEqual(new double[0], result.MultiNumericColumns[0][2]);
CollectionAssert.AreEqual(new [] { 3.0 }, result.MultiNumericColumns[0][3]);
CollectionAssert.AreEqual(new [] { 0.0, 1.0 }, result.MultiNumericColumns[1][0]);
CollectionAssert.AreEqual(new [] { 2.0 }, result.MultiNumericColumns[1][1]);
CollectionAssert.AreEqual(new double[0], result.MultiNumericColumns[1][2]);
CollectionAssert.AreEqual(new [] { 3.0 }, result.MultiNumericColumns[1][3]);
}
