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]);
}
public static void Read(INetworkData ndata, string outFolder, ProcessInfo processInfo)
{
ReadMatrixDataInto(ndata, Path.Combine(outFolder, "networks.txt"), processInfo);
foreach (var netAttr in ndata.GetStringColumn("guid").Zip(ndata.GetStringColumn("name"), (guid, name) => new { guid, name }))
{
var guid = Guid.Parse(netAttr.guid);
var nodeTable = PerseusFactory.CreateDataWithAnnotationColumns();
var edgeTable = PerseusFactory.CreateDataWithAnnotationColumns();
ReadMatrixDataInto(nodeTable, Path.Combine(outFolder, $"{guid}_nodes.txt"), processInfo);
ReadMatrixDataInto(edgeTable, Path.Combine(outFolder, $"{guid}_edges.txt"), processInfo);
var graph = new Graph();
var nodeIndex = new Dictionary <INode, int>();
var nameToNode = new Dictionary <string, INode>();
var nodeColumn = nodeTable.GetStringColumn("node");
for (int row = 0; row < nodeTable.RowCount; row++)
{
var node = graph.AddNode();
nodeIndex[node] = row;
nameToNode[nodeColumn[row]] = node;
}
var sourceColumn = edgeTable.GetStringColumn("source");
var targetColumn = edgeTable.GetStringColumn("target");
var edgeIndex = new Dictionary <IEdge, int>();
for (int row = 0; row < edgeTable.RowCount; row++)
{
var source = nameToNode[sourceColumn[row]];
var target = nameToNode[targetColumn[row]];
var edge = graph.AddEdge(source, target);
edgeIndex[edge] = row;
}
ndata.AddNetworks(new NetworkInfo(graph, nodeTable, nodeIndex, edgeTable, edgeIndex, netAttr.name, guid));
}
ReadMatrixDataInto(ndata, Path.Combine(outFolder, "networks.txt"), processInfo);
}
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 TestWriteMultiNumericColumnWithNulls()
{
var data = PerseusFactory.CreateDataWithAnnotationColumns();
data.AddMultiNumericColumn("Test", "", new double[1][]);
data.AddStringColumn("Test2", "", new string[1]);
Assert.AreEqual(1, data.RowCount);
var writer = new StreamWriter(new MemoryStream());
PerseusUtils.WriteDataWithAnnotationColumns(data, writer);
}
public void TestNetworkUniqueRows()
{
Random RandGen = new Random();
INetworkData ndata = new NetworkData();
ndata.Name = "Random network(s)";
ndata.Description = ndata.Name;
var n = 3;
var numNodes = 100;
var numEdges = 150;
for (int i = 0; i < n; i++)
{
var graph = new Graph();//!!!!
var nodeTable = PerseusFactory.CreateDataWithAnnotationColumns();
var nodeIndex = new Dictionary <INode, int>();
var edgeTable = PerseusFactory.CreateDataWithAnnotationColumns();
var edgeIndex = new Dictionary <IEdge, int>();
for (int j = 0; j < numNodes; j++)
{
nodeIndex[graph.AddNode()] = j;
}
var nodeNames = Enumerable.Range(0, graph.NumberOfNodes).Select(x => $"node {x}").ToArray();
nodeTable.AddStringColumn("Node", "", nodeNames);
var nodes = graph.ToArray();
var sources = new List <string>();
var targets = new List <string>();
for (int j = 0; j < numEdges; j++)
{
var source = nodes[RandGen.Next(0, nodes.Length)];
sources.Add(nodeNames[nodeIndex[source]]);
var target = nodes[RandGen.Next(0, nodes.Length)];
targets.Add(nodeNames[nodeIndex[target]]);
edgeIndex[graph.AddEdge(source, target)] = j;
}
edgeTable.AddStringColumn("Source", "", sources.ToArray());
edgeTable.AddStringColumn("Target", "", targets.ToArray());
var network = new NetworkInfo(graph, nodeTable, nodeIndex, edgeTable, edgeIndex, $"Random {i}");
ndata.AddNetworks(network);
}
foreach (var network in ndata)
{
network.UniqueRows(network.EdgeTable.StringColumns[0], network.EdgeTable.StringColumns[1], network.NodeTable.StringColumns[0], ArrayUtils.Median, RemoveDuplicateEdges.Union, RemoveDuplicateEdges.CatUnion, RemoveDuplicateEdges.MultiNumUnion);
Assert.True(network.EdgeTable.RowCount <= 150);
Assert.True(network.EdgeTable.RowCount == network.Graph.NumberOfEdges);
Assert.True(network.NodeTable.RowCount == network.Graph.NumberOfNodes);
}
}