private static string ProcessDataReadFromFile(IDataWithAnnotationRows mdata, Parameters param)
{
Parameter <string> fp = param.GetParam <string>("Input file");
string filename = fp.Value;
string[] colNames;
try {
colNames = TabSep.GetColumnNames(filename, '\t');
} catch (Exception) {
return("Could not open file " + filename + ". It maybe open in another program.");
}
int nameIndex = GetNameIndex(colNames);
if (nameIndex < 0)
{
return("Error: the file has to contain a column called 'Name'.");
}
if (colNames.Length < 2)
{
return("Error: the file does not contain a numerical column.");
}
string[] nameCol = TabSep.GetColumn(colNames[nameIndex], filename, '\t');
Dictionary <string, int> map = ArrayUtils.InverseMap(nameCol);
for (int i = 0; i < colNames.Length; i++)
{
if (i == nameIndex)
{
continue;
}
string groupName = colNames[i];
string[] groupCol = TabSep.GetColumn(groupName, filename, '\t');
double[] newCol = new double[mdata.ColumnCount];
for (int j = 0; j < newCol.Length; j++)
{
string colName = mdata.ColumnNames[j];
if (!map.ContainsKey(colName))
{
newCol[j] = double.NaN;
continue;
}
int ind = map[colName];
string group = groupCol[ind] ?? "";
group = group.Trim();
if (string.IsNullOrEmpty(group))
{
newCol[j] = double.NaN;
}
else
{
if (!Parser.TryDouble(group, out newCol[j]))
{
newCol[j] = double.NaN;
}
}
}
mdata.AddNumericRow(groupName, groupName, newCol);
}
return(null);
}
private static string ProcessDataReadFromFile(IDataWithAnnotationRows mdata, Parameters param)
{
Parameter <string> fp = param.GetParam <string>("Input file");
string filename = fp.Value;
string[] colNames;
try
{
colNames = TabSep.GetColumnNames(filename, '\t');
}
catch (Exception)
{
return("Could not open file " + filename + ". It maybe open in another program.");
}
int nameIndex = GetNameIndex(colNames);
if (nameIndex < 0)
{
return("Error: the file has to contain a column called 'Name'.");
}
if (colNames.Length < 2)
{
return("Error: the file does not contain a grouping column.");
}
string[] nameCol = TabSep.GetColumn(colNames[nameIndex], filename, '\t');
for (int i = 0; i < nameCol.Length; i++)
{
MessageBox.Show(nameCol[i].ToString());
}
return(null);
}
private static string ProcessDataReadFromFileList(IMatrixData mdata, Parameters param)
{
Parameter <string> fp = param.GetParam <string>("Input file");
string filename = fp.Value;
string[] colNames;
try
{
colNames = TabSep.GetColumnNames(filename, '\t');
}
catch (Exception)
{
return("Could not open file " + filename + ". It maybe open in another program.");
}
int nameIndex = GetNameIndex(colNames);
string[] nameCol = TabSep.GetColumn(colNames[nameIndex], filename, '\t');
MessageBox.Show(nameCol.Length.ToString());
MessageBox.Show(mdata.ColumnCount.ToString());
for (int i = 0; i < mdata.ColumnCount; i++)
{
MessageBox.Show(nameCol.Length.ToString());
MessageBox.Show(mdata.ColumnCount.ToString());
/* if (nameCol.Length != mdata.ColumnCount)
* {
* return "ERROR";
* }
* else { mdata.ColumnNames[i] = nameCol[i]; } */
}
return(null);
}
private static string ProcessDataReadFromFile(IMatrixData mdata, Parameters param)
{
Parameter <string> fp = param.GetParam <string>("Input file");
string filename = fp.Value;
string[] colNames = TabSep.GetColumnNames(filename, '\t');
int nameIndex = GetNameIndex(colNames);
if (nameIndex < 0)
{
return("Error: the file has to contain a column called 'Name'.");
}
if (colNames.Length < 2)
{
return("Error: the file does not contain a grouping column.");
}
string[] nameCol = TabSep.GetColumn(colNames[nameIndex], filename, '\t');
Dictionary <string, int> map = ArrayUtils.InverseMap(nameCol);
for (int i = 0; i < colNames.Length; i++)
{
if (i == nameIndex)
{
continue;
}
string groupName = colNames[i];
string[] groupCol = TabSep.GetColumn(groupName, filename, '\t');
string[][] newCol = new string[mdata.ColumnCount][];
for (int j = 0; j < newCol.Length; j++)
{
string colName = mdata.ColumnNames[j];
if (!map.ContainsKey(colName))
{
newCol[j] = new string[0];
continue;
}
int ind = map[colName];
string group = groupCol[ind] ?? "";
group = group.Trim();
if (string.IsNullOrEmpty(group))
{
newCol[j] = new string[0];
}
else
{
string[] w = group.Split(';');
Array.Sort(w);
for (int k = 0; k < w.Length; k++)
{
w[k] = w[k].Trim();
}
newCol[j] = w;
}
}
mdata.AddCategoryRow(groupName, groupName, newCol);
}
return(null);
}
public static void ParseKinaseSubstrate(string filename, out string[] seqWins, out string[] subAccs,
out string[] kinases, out string[] kinAccs, out string[] species)
{
seqWins = TabSep.GetColumn("SITE_+/-7_AA", filename, 3, '\t');
subAccs = TabSep.GetColumn("SUB_ACC_ID", filename, 3, '\t');
kinases = TabSep.GetColumn("KINASE", filename, 3, '\t');
kinAccs = TabSep.GetColumn("KIN_ACC_ID", filename, 3, '\t');
species = TabSep.GetColumn("KIN_ORGANISM", filename, 3, '\t');
}
public static void ParseKnownMods(string filename, out string[] seqWins, out string[] accs, out string[] pubmedLtp,
out string[] pubmedMs2, out string[] cstMs2, out string[] species)
{
seqWins = TabSep.GetColumn("SITE_+/-7_AA", filename, 3, '\t');
accs = TabSep.GetColumn("ACC_ID", filename, 3, '\t');
pubmedLtp = TabSep.GetColumn("LT_LIT", filename, 3, '\t');
pubmedMs2 = TabSep.GetColumn("MS_LIT", filename, 3, '\t');
cstMs2 = TabSep.GetColumn("MS_CST", filename, 3, '\t');
species = TabSep.GetColumn("ORGANISM", filename, 3, '\t');
}
public static string[] GetAllKinaseSubstrateOrganisms()
{
string filename = GetKinaseSubstrateFile();
if (filename == null)
{
return(null);
}
string[] species = TabSep.GetColumn("KIN_ORGANISM", filename, 3, '\t');
return(ArrayUtils.UniqueValues(species));
}
public static void ParseRegulatorySites(string filename, out string[] seqWins, out string[] accs,
out string[] function, out string[] process, out string[] protInteract, out string[] otherInteract,
out string[] notes, out string[] species)
{
seqWins = TabSep.GetColumn("SITE_+/-7_AA", filename, 3, '\t');
accs = TabSep.GetColumn("ACC_ID", filename, 3, '\t');
function = TabSep.GetColumn("ON_FUNCTION", filename, 3, '\t');
process = TabSep.GetColumn("ON_PROCESS", filename, 3, '\t');
protInteract = TabSep.GetColumn("ON_PROT_INTERACT", filename, 3, '\t');
otherInteract = TabSep.GetColumn("ON_OTHER_INTERACT", filename, 3, '\t');
notes = TabSep.GetColumn("NOTES", filename, 3, '\t');
species = TabSep.GetColumn("ORGANISM", filename, 3, '\t');
}
private static string ProcessDataReadFromFileList(IMatrixData mdata, Parameters param)
{
Parameter <string> fp = param.GetParam <string>("Input file");
string filename = fp.Value;
string[] colNames;
try{
colNames = TabSep.GetColumnNames(filename, '\t');
} catch (Exception) {
return("Could not open file " + filename + ". It maybe open in another program.");
}
int nameIndex = GetNameIndex(colNames);
string[] nameCol = TabSep.GetColumn(colNames[nameIndex], filename, '\t');
return(null);
}
public void LoadData(IMatrixData matrixData, Parameters parameters, ref IMatrixData[] supplTables,
ref IDocumentData[] documents, ProcessInfo processInfo)
{
int ind = parameters.GetParam <int>("Organism").Value;
string filename = GetOrganismFiles()[ind];
List <string> stringColnames = new List <string>(TabSep.GetColumnNames(filename, '\t'));
List <string[]> stringCols = new List <string[]>();
foreach (string t in stringColnames)
{
string[] col = TabSep.GetColumn(t, filename, '\t');
stringCols.Add(col);
}
matrixData.Name = "Gene list";
matrixData.ColumnNames = new List <string>();
matrixData.Values.Init(stringCols[0].Length, 0);
matrixData.SetAnnotationColumns(stringColnames, stringCols,
new List <string>(), new List <string[][]>(), new List <string>(), new List <double[]>(), new List <string>(),
new List <double[][]>());
matrixData.Origin = "Gene list";
}
