/// <summary>
/// Main routine that opens data files (Expressions, Factors)
/// in comma or tab delimited CSVs (and Excel as a future possibility).
/// The main file load routine is in clsDataTable.LoadFile2DataTable(fileName)
/// Duplicates are handled in C#
/// </summary>
private bool OpenFile(string filePath)
{
string rowID;
string rcmd;
var success = true;
bool validFactors;
// check that the file exists before opening it
if (!File.Exists(filePath))
{
return(false);
}
if (!string.Equals(mstrLoadedfileName, filePath))
{
mstrLoadedfileName = filePath;
}
var fExt = Path.GetExtension(filePath);
if (string.IsNullOrWhiteSpace(fExt))
{
MessageBox.Show("File does not have an extension (like .csv or .txt); unable to determine file type",
"Error!", MessageBoxButtons.OK, MessageBoxIcon.Exclamation);
return(false);
}
if (!ValidExtension(fExt))
{
MessageBox.Show("Filetype not allowed (must be csv, txt, xls, or xlsx)", "Error!");
return(false);
}
var dataLoader = new clsDataTable();
dataLoader.OnError += clsDataTable_OnError;
dataLoader.OnProgress += clsDataTable_OnProgress;
switch (mDataSetType)
{
case enmDataType.ESET:
#region Load Expressions
validFactors = true;
mfrmShowProgress.Reset("Loading data");
var esetTable = dataLoader.LoadFile2DataTableFastCSVReader(mstrLoadedfileName);
if (esetTable == null)
{
string errorMessage;
if (string.IsNullOrWhiteSpace(mfrmShowProgress.ErrorMessage))
{
errorMessage = "Unknown load error";
}
else
{
errorMessage = "Load error: " + mfrmShowProgress.ErrorMessage;
}
MessageBox.Show(errorMessage, "Error", MessageBoxButtons.OK, MessageBoxIcon.Exclamation);
return(false);
}
esetTable.TableName = "AllEset";
//Select columns
var columnSelectionForm = new frmSelectColumns
{
PopulateListBox = clsDataTable.DataTableColumns(esetTable, false),
Proteins = mhtDatasets.ContainsKey("Protein Info")
};
if (columnSelectionForm.ShowDialog() == DialogResult.OK)
{
rowID = columnSelectionForm.RowIDColumn; //mass tags
var dataCols = columnSelectionForm.DataColumns.ToList();
try
{
var filteredDataTable = ArrangeDataTable(esetTable, rowID, dataCols);
// Rename the first column from MassTagID (or whatever the user-supplied name is) to Row_ID
filteredDataTable.Columns[0].ColumnName = "Row_ID";
// Remove rows with no data or duplicate data
filteredDataTable = dataLoader.RemoveDuplicateRows2(
filteredDataTable,
filteredDataTable.Columns[0].ColumnName);
// Copy the data into R
filteredDataTable.TableName = "Eset";
success = mRConnector.SendTable2RmatrixNumeric("Eset", filteredDataTable);
if (mhtDatasets.ContainsKey("Factors"))
{
rcmd = "FactorsValid<-identical(as.array(colnames(factors)),as.array(colnames(Eset)))";
mRConnector.EvaluateNoReturn(rcmd);
validFactors = mRConnector.GetSymbolAsBool("FactorsValid");
}
if (!validFactors)
{
success = false;
}
if (success)
{
AddDataset2HashTable(filteredDataTable);
mRConnector.EvaluateNoReturn("print(dim(Eset))");
mRConnector.EvaluateNoReturn("cat(\"Expressions loaded.\n\")");
}
}
catch (Exception ex)
{
MessageBox.Show("Error: " + ex.Message,
"Exception while extracting data columns");
return(false);
}
if (columnSelectionForm.Proteins &&
!string.IsNullOrWhiteSpace(columnSelectionForm.ProteinIDColumn) && success)
{
// Load protein info then send to R
var proteinDataTable = LoadProtColumns(esetTable, columnSelectionForm.ProteinIDColumn, rowID,
columnSelectionForm.ProteinMetadataColumns);
proteinDataTable.TableName = "ProtInfo";
AddDataset2HashTable(proteinDataTable);
}
}
else
{
success = false;
}
#endregion
break;
case enmDataType.PROTINFO:
#region Load Protein Info
var proteinInfoTable = dataLoader.LoadFile2DataTableFastCSVReader(mstrLoadedfileName);
if (proteinInfoTable == null)
{
return(false);
}
var proteinSelectionForm = new frmSelectProtInfo
{
PopulateListBox = clsDataTable.DataTableColumns(proteinInfoTable, false)
};
if (proteinSelectionForm.ShowDialog() == DialogResult.OK)
{
rowID = proteinSelectionForm.RowIDColumn; //mass tags
var proteinIdentifierColumn = proteinSelectionForm.ProteinIDColumn;
var proteinDataTable = LoadProtColumns(proteinInfoTable, proteinIdentifierColumn, rowID,
proteinSelectionForm.ProteinMetadataColumns);
proteinDataTable.TableName = "ProtInfo";
AddDataset2HashTable(proteinDataTable);
}
else
{
success = false;
}
#endregion
break;
case enmDataType.FACTORS:
// Factor files can be CSV files or .txt files
// The first row of the factor definitions file must have a column named Factor, then column names that match the names in the originally loaded Expressions table
// Each subsequent row of the factor file is a new factor name, then the factor value for each dataset
//
// Example factor file (.txt file should use a tab delimiter; CSV file would use commas)
//
// Factor P10A P10B P11A P11B P12A P12B
// Time 0 0 5 5 10 10
// Temperature Hot Cold Hot Cold Hot Cold
#region Load Factors
validFactors = true;
var factorTable = dataLoader.LoadFile2DataTableFastCSVReader(mstrLoadedfileName);
if (factorTable == null)
{
return(false);
}
if (factorTable.Rows.Count > frmDefFactors.MAX_LEVELS)
{
MessageBox.Show(
"Factors file has too many factors; max allowed is " + frmDefFactors.MAX_LEVELS + " factors",
"Error", MessageBoxButtons.OK, MessageBoxIcon.Exclamation);
return(false);
}
if (mRConnector.SendTable2RmatrixNonNumeric("factors", factorTable))
{
try
{
if (mhtDatasets.ContainsKey("Expressions"))
{
rcmd = "FactorsValid<-identical(as.array(colnames(factors)),as.array(colnames(Eset)))";
mRConnector.EvaluateNoReturn(rcmd);
validFactors = mRConnector.GetSymbolAsBool("FactorsValid");
}
if (!validFactors)
{
success = false;
}
else
{
UpdateFactorInfoArray();
factorTable.Columns[0].ColumnName = "Factors";
factorTable.TableName = "factors";
mRConnector.EvaluateNoReturn("print(factors)");
mRConnector.EvaluateNoReturn("cat(\"Factors loaded.\n\")");
}
}
catch (Exception ex)
{
MessageBox.Show("Error: " + ex.Message, "Exception while talking to R");
success = false;
}
}
if (success)
{
AddDataset2HashTable(factorTable);
}
#endregion
break;
}
return(success);
}
private bool OpenFile2(string filename)
{
var success = true;
// check that the file exists before opening it
if (!File.Exists(mstrLoadedfileName))
{
return(false);
}
var fExt = Path.GetExtension(filename);
if (!ValidExtension(fExt))
{
MessageBox.Show("Filetype not allowed.", "Error!");
return(false);
}
string rowID;
string rcmd;
bool FactorsValid;
switch (mDataSetType)
{
case enmDataType.ESET:
#region Load Expressions
FactorsValid = true;
var mDTloaded = clsDataTable.LoadFile2DataTableFastCSVReader(mstrLoadedfileName);
if (mDTloaded == null)
{
return(false);
}
mDTloaded.TableName = "AllEset";
//Select columns
var mfrmSelectCols = new frmSelectColumns
{
PopulateListBox = clsDataTable.DataTableColumns(mDTloaded, false),
Proteins = mhtDatasets.ContainsKey("Protein Info")
};
if (mfrmSelectCols.ShowDialog() == DialogResult.OK)
{
rowID = mfrmSelectCols.RowIDColumn; //mass tags
var dataCols = mfrmSelectCols.DataColumns;
DataTable mDTselectedEset1;
try
{
mDTselectedEset1 = ArrangeDataTable(mDTloaded, rowID, dataCols); // create the datatable
mDTselectedEset1.TableName = "Eset";
}
catch (Exception ex)
{
MessageBox.Show("Error: " + ex.Message,
"Exception while extracting data columns");
return(false);
}
//clsRarray.rowNamesID = mDTselectedEset.Columns[0].ToString();
clsRarray.rowNamesID = "Row_ID";
if (mRConnector.SendTable2RmatrixNumeric("Eset", mDTselectedEset1))
// Duplicates are handled during the call 'SendTable2RmatrixNumeric'
{
try
{
if (mhtDatasets.ContainsKey("Factors"))
{
rcmd = "FactorsValid<-identical(as.array(colnames(factors)),as.array(colnames(Eset)))";
mRConnector.EvaluateNoReturn(rcmd);
FactorsValid = mRConnector.GetSymbolAsBool("FactorsValid");
}
if (!FactorsValid)
{
success = false;
}
else
{
mRConnector.GetTableFromRmatrix("Eset"); // Get the cleaned data matrix
mDTselectedEset1 = mRConnector.DataTable.Copy();
mDTselectedEset1.TableName = "Eset";
mRConnector.EvaluateNoReturn("print(dim(Eset))");
mRConnector.EvaluateNoReturn("cat(\"Expressions loaded.\n\")");
AddDataset2HashTable(mDTselectedEset1);
}
}
catch (Exception ex)
{
MessageBox.Show("Error: " + ex.Message, "Exception while talking to R");
success = false;
}
}
else
{
success = false;
}
if (mfrmSelectCols.Proteins && success) // Protein info needs to be loaded ?
{
// loads to mDataTableProtInfo and then sends to R
var mdtProts = LoadProtColumns(mDTloaded, mfrmSelectCols.ProteinIDColumn, rowID);
mdtProts.TableName = "ProtInfo";
AddDataset2HashTable(mdtProts);
}
}
else
{
success = false;
}
#endregion
break;
case enmDataType.PROTINFO:
#region Load Protein Info
var mDTtmp = clsDataTable.LoadFile2DataTableFastCSVReader(mstrLoadedfileName);
if (mDTtmp == null)
{
return(false);
}
var mfrmSelectProts = new frmSelectProtInfo
{
PopulateListBox = clsDataTable.DataTableColumns(mDTtmp, false)
};
if (mfrmSelectProts.ShowDialog() == DialogResult.OK)
{
rowID = mfrmSelectProts.RowIDColumn; //mass tags
var protIPI = mfrmSelectProts.ProteinIDColumn;
var mdtProts = LoadProtColumns(mDTtmp, protIPI, rowID);
mdtProts.TableName = "ProtInfo";
AddDataset2HashTable(mdtProts);
}
else
{
success = false;
}
#endregion
break;
case enmDataType.FACTORS:
#region Load Factors
FactorsValid = true;
var mDTFactors = clsDataTable.LoadFile2DataTableFastCSVReader(mstrLoadedfileName);
if (mDTFactors == null)
{
return(false);
}
if (mDTFactors.Rows.Count > frmDefFactors.MAX_LEVELS)
{
MessageBox.Show("Factors file has too many factors; max allowed is " + frmDefFactors.MAX_LEVELS + " factors", "Error", MessageBoxButtons.OK, MessageBoxIcon.Exclamation);
return(false);
}
if (mRConnector.SendTable2RmatrixNonNumeric("factors", mDTFactors))
{
try
{
if (mhtDatasets.ContainsKey("Expressions"))
{
rcmd = "FactorsValid<-identical(as.array(colnames(factors)),as.array(colnames(Eset)))";
mRConnector.EvaluateNoReturn(rcmd);
FactorsValid = mRConnector.GetSymbolAsBool("FactorsValid");
}
if (!FactorsValid)
{
success = false;
}
else
{
UpdateFactorInfoArray();
mDTFactors.Columns[0].ColumnName = "Factors";
mDTFactors.TableName = "factors";
mRConnector.EvaluateNoReturn("print(factors)");
mRConnector.EvaluateNoReturn("cat(\"Factors loaded.\n\")");
}
}
catch (Exception ex)
{
MessageBox.Show("Error: " + ex.Message, "Exception while talking to R");
success = false;
}
}
if (success)
{
AddDataset2HashTable(mDTFactors);
}
#endregion
break;
default:
break;
}
return(success);
}
/// <summary>
/// Main routine that opens data files (Expressions, Factors)
/// in comma or tab delimited CSVs (and Excel as a future possibility).
/// The main file load routine is in clsDataTable.LoadFile2DataTable(fileName)
/// Duplicates are handled in C#
/// </summary>
private bool OpenFile(string filePath)
{
string rowID;
string rcmd;
var success = true;
bool FactorsValid;
// check that the file exists before opening it
if (!File.Exists(filePath))
{
return(false);
}
if (!string.Equals(mstrLoadedfileName, filePath))
{
mstrLoadedfileName = filePath;
}
var fExt = Path.GetExtension(filePath);
if (string.IsNullOrWhiteSpace(fExt))
{
MessageBox.Show("File does not have an extension (like .csv or .txt); unable to determine file type", "Error!", MessageBoxButtons.OK, MessageBoxIcon.Exclamation);
return(false);
}
if (!ValidExtension(fExt))
{
MessageBox.Show("Filetype not allowed (must be csv, txt, xls, or xlsx)", "Error!");
return(false);
}
switch (mDataSetType)
{
case enmDataType.ESET:
#region Load Expressions
FactorsValid = true;
if (!mProgressEventWired)
{
clsDataTable.OnProgress += clsDataTable_OnProgress;
mProgressEventWired = true;
}
var mDTloaded = clsDataTable.LoadFile2DataTableFastCSVReader(mstrLoadedfileName);
if (mDTloaded == null)
{
return(false);
}
mDTloaded.TableName = "AllEset";
//Select columns
var mfrmSelectCols = new frmSelectColumns
{
PopulateListBox = clsDataTable.DataTableColumns(mDTloaded, false),
Proteins = mhtDatasets.ContainsKey("Protein Info")
};
if (mfrmSelectCols.ShowDialog() == DialogResult.OK)
{
rowID = mfrmSelectCols.RowIDColumn; //mass tags
var dataCols = mfrmSelectCols.DataColumns;
try
{
var mDTselectedEset1 = ArrangeDataTable(mDTloaded, rowID, dataCols);
mDTselectedEset1.Columns[0].ColumnName = "Row_ID";
mDTselectedEset1 = clsDataTable.RemoveDuplicateRows2(mDTselectedEset1,
mDTselectedEset1.Columns[0].ColumnName);
// handle duplicate rows
mDTselectedEset1.TableName = "Eset";
success = mRConnector.SendTable2RmatrixNumeric("Eset", mDTselectedEset1);
if (mhtDatasets.ContainsKey("Factors"))
{
rcmd = "FactorsValid<-identical(as.array(colnames(factors)),as.array(colnames(Eset)))";
mRConnector.EvaluateNoReturn(rcmd);
FactorsValid = mRConnector.GetSymbolAsBool("FactorsValid");
}
if (!FactorsValid)
{
success = false;
}
if (success)
{
AddDataset2HashTable(mDTselectedEset1);
mRConnector.EvaluateNoReturn("print(dim(Eset))");
mRConnector.EvaluateNoReturn("cat(\"Expressions loaded.\n\")");
}
}
catch (Exception ex)
{
MessageBox.Show("Error: " + ex.Message,
"Exception while extracting data columns");
return(false);
}
if (mfrmSelectCols.Proteins && success) // Protein info needs to be loaded ?
{
// loads to mDataTableProtInfo and then sends to R
var mdtProts = LoadProtColumns(mDTloaded, mfrmSelectCols.ProteinIDColumn, rowID);
mdtProts.TableName = "ProtInfo";
AddDataset2HashTable(mdtProts);
}
}
else
{
success = false;
}
#endregion
break;
case enmDataType.PROTINFO:
#region Load Protein Info
var mDTtmp = clsDataTable.LoadFile2DataTableFastCSVReader(mstrLoadedfileName);
if (mDTtmp == null)
{
return(false);
}
var mfrmSelectProts = new frmSelectProtInfo
{
PopulateListBox = clsDataTable.DataTableColumns(mDTtmp, false)
};
if (mfrmSelectProts.ShowDialog() == DialogResult.OK)
{
rowID = mfrmSelectProts.RowIDColumn; //mass tags
var protIPI = mfrmSelectProts.ProteinIDColumn;
var mdtProts = LoadProtColumns(mDTtmp, protIPI, rowID);
mdtProts.TableName = "ProtInfo";
AddDataset2HashTable(mdtProts);
}
else
{
success = false;
}
#endregion
break;
case enmDataType.FACTORS:
// Factor files can be CSV files or .txt files
// The first row of the factor definitions file must have a column named Factor, then column names that match the names in the originally loaded Expressions table
// Each subsequent row of the factor file is a new factor name, then the factor value for each dataset
//
// Example factor file (.txt file should use a tab delimiter; CSV file would use commas)
//
// Factor P10A P10B P11A P11B P12A P12B
// Time 0 0 5 5 10 10
// Temperature Hot Cold Hot Cold Hot Cold
#region Load Factors
FactorsValid = true;
var mDTFactors = clsDataTable.LoadFile2DataTableFastCSVReader(mstrLoadedfileName);
if (mDTFactors == null)
{
return(false);
}
if (mDTFactors.Rows.Count > frmDefFactors.MAX_LEVELS)
{
MessageBox.Show("Factors file has too many factors; max allowed is " + frmDefFactors.MAX_LEVELS + " factors", "Error", MessageBoxButtons.OK, MessageBoxIcon.Exclamation);
return(false);
}
if (mRConnector.SendTable2RmatrixNonNumeric("factors", mDTFactors))
{
try
{
if (mhtDatasets.ContainsKey("Expressions"))
{
rcmd = "FactorsValid<-identical(as.array(colnames(factors)),as.array(colnames(Eset)))";
mRConnector.EvaluateNoReturn(rcmd);
FactorsValid = mRConnector.GetSymbolAsBool("FactorsValid");
}
if (!FactorsValid)
{
success = false;
}
else
{
UpdateFactorInfoArray();
mDTFactors.Columns[0].ColumnName = "Factors";
mDTFactors.TableName = "factors";
mRConnector.EvaluateNoReturn("print(factors)");
mRConnector.EvaluateNoReturn("cat(\"Factors loaded.\n\")");
}
}
catch (Exception ex)
{
MessageBox.Show("Error: " + ex.Message, "Exception while talking to R");
success = false;
}
}
if (success)
{
AddDataset2HashTable(mDTFactors);
}
#endregion
break;
default:
break;
}
return(success);
}
private bool OpenFile2(string filename)
{
var success = true;
// check that the file exists before opening it
if (!File.Exists(mstrLoadedfileName))
{
return false;
}
var fExt = Path.GetExtension(filename);
if (!ValidExtension(fExt))
{
MessageBox.Show("Filetype not allowed.", "Error!");
return false;
}
string rowID;
string rcmd;
bool FactorsValid;
switch (mDataSetType)
{
case enmDataType.ESET:
#region Load Expressions
FactorsValid = true;
var mDTloaded = clsDataTable.LoadFile2DataTableFastCSVReader(mstrLoadedfileName);
if (mDTloaded == null)
{
return false;
}
mDTloaded.TableName = "AllEset";
//Select columns
var mfrmSelectCols = new frmSelectColumns
{
PopulateListBox = clsDataTable.DataTableColumns(mDTloaded, false),
Proteins = mhtDatasets.ContainsKey("Protein Info")
};
if (mfrmSelectCols.ShowDialog() == DialogResult.OK)
{
rowID = mfrmSelectCols.RowIDColumn; //mass tags
var dataCols = mfrmSelectCols.DataColumns;
DataTable mDTselectedEset1;
try
{
mDTselectedEset1 = ArrangeDataTable(mDTloaded, rowID, dataCols); // create the datatable
mDTselectedEset1.TableName = "Eset";
}
catch (Exception ex)
{
MessageBox.Show("Error: " + ex.Message,
"Exception while extracting data columns");
return false;
}
//clsRarray.rowNamesID = mDTselectedEset.Columns[0].ToString();
clsRarray.rowNamesID = "Row_ID";
if (mRConnector.SendTable2RmatrixNumeric("Eset", mDTselectedEset1))
// Duplicates are handled during the call 'SendTable2RmatrixNumeric'
{
try
{
if (mhtDatasets.ContainsKey("Factors"))
{
rcmd = "FactorsValid<-identical(as.array(colnames(factors)),as.array(colnames(Eset)))";
mRConnector.EvaluateNoReturn(rcmd);
FactorsValid = mRConnector.GetSymbolAsBool("FactorsValid");
}
if (!FactorsValid)
{
success = false;
}
else
{
mRConnector.GetTableFromRmatrix("Eset"); // Get the cleaned data matrix
mDTselectedEset1 = mRConnector.DataTable.Copy();
mDTselectedEset1.TableName = "Eset";
mRConnector.EvaluateNoReturn("print(dim(Eset))");
mRConnector.EvaluateNoReturn("cat(\"Expressions loaded.\n\")");
AddDataset2HashTable(mDTselectedEset1);
}
}
catch (Exception ex)
{
MessageBox.Show("Error: " + ex.Message, "Exception while talking to R");
success = false;
}
}
else
{
success = false;
}
if (mfrmSelectCols.Proteins && success) // Protein info needs to be loaded ?
{
// loads to mDataTableProtInfo and then sends to R
var mdtProts = LoadProtColumns(mDTloaded, mfrmSelectCols.ProteinIDColumn, rowID);
mdtProts.TableName = "ProtInfo";
AddDataset2HashTable(mdtProts);
}
}
else
{
success = false;
}
#endregion
break;
case enmDataType.PROTINFO:
#region Load Protein Info
var mDTtmp = clsDataTable.LoadFile2DataTableFastCSVReader(mstrLoadedfileName);
if (mDTtmp == null)
{
return false;
}
var mfrmSelectProts = new frmSelectProtInfo
{
PopulateListBox = clsDataTable.DataTableColumns(mDTtmp, false)
};
if (mfrmSelectProts.ShowDialog() == DialogResult.OK)
{
rowID = mfrmSelectProts.RowIDColumn; //mass tags
var protIPI = mfrmSelectProts.ProteinIDColumn;
var mdtProts = LoadProtColumns(mDTtmp, protIPI, rowID);
mdtProts.TableName = "ProtInfo";
AddDataset2HashTable(mdtProts);
}
else
{
success = false;
}
#endregion
break;
case enmDataType.FACTORS:
#region Load Factors
FactorsValid = true;
var mDTFactors = clsDataTable.LoadFile2DataTableFastCSVReader(mstrLoadedfileName);
if (mDTFactors == null)
{
return false;
}
if (mDTFactors.Rows.Count > frmDefFactors.MAX_LEVELS)
{
MessageBox.Show("Factors file has too many factors; max allowed is " + frmDefFactors.MAX_LEVELS + " factors", "Error", MessageBoxButtons.OK, MessageBoxIcon.Exclamation);
return false;
}
if (mRConnector.SendTable2RmatrixNonNumeric("factors", mDTFactors))
{
try
{
if (mhtDatasets.ContainsKey("Expressions"))
{
rcmd = "FactorsValid<-identical(as.array(colnames(factors)),as.array(colnames(Eset)))";
mRConnector.EvaluateNoReturn(rcmd);
FactorsValid = mRConnector.GetSymbolAsBool("FactorsValid");
}
if (!FactorsValid)
{
success = false;
}
else
{
UpdateFactorInfoArray();
mDTFactors.Columns[0].ColumnName = "Factors";
mDTFactors.TableName = "factors";
mRConnector.EvaluateNoReturn("print(factors)");
mRConnector.EvaluateNoReturn("cat(\"Factors loaded.\n\")");
}
}
catch (Exception ex)
{
MessageBox.Show("Error: " + ex.Message, "Exception while talking to R");
success = false;
}
}
if (success)
{
AddDataset2HashTable(mDTFactors);
}
#endregion
break;
default:
break;
}
return success;
}
/// <summary>
/// Main routine that opens data files (Expressions, Factors)
/// in comma or tab delimited CSVs (and Excel as a future possibility).
/// The main file load routine is in clsDataTable.LoadFile2DataTable(fileName)
/// Duplicates are handled in C#
/// </summary>
private bool OpenFile(string filePath)
{
string rowID;
string rcmd;
var success = true;
bool FactorsValid;
// check that the file exists before opening it
if (!File.Exists(filePath))
{
return false;
}
if (!string.Equals(mstrLoadedfileName, filePath))
{
mstrLoadedfileName = filePath;
}
var fExt = Path.GetExtension(filePath);
if (string.IsNullOrWhiteSpace(fExt))
{
MessageBox.Show("File does not have an extension (like .csv or .txt); unable to determine file type", "Error!", MessageBoxButtons.OK, MessageBoxIcon.Exclamation);
return false;
}
if (!ValidExtension(fExt))
{
MessageBox.Show("Filetype not allowed (must be csv, txt, xls, or xlsx)", "Error!");
return false;
}
switch (mDataSetType)
{
case enmDataType.ESET:
#region Load Expressions
FactorsValid = true;
if (!mProgressEventWired)
{
clsDataTable.OnProgress += clsDataTable_OnProgress;
mProgressEventWired = true;
}
var mDTloaded = clsDataTable.LoadFile2DataTableFastCSVReader(mstrLoadedfileName);
if (mDTloaded == null)
{
return false;
}
mDTloaded.TableName = "AllEset";
//Select columns
var mfrmSelectCols = new frmSelectColumns
{
PopulateListBox = clsDataTable.DataTableColumns(mDTloaded, false),
Proteins = mhtDatasets.ContainsKey("Protein Info")
};
if (mfrmSelectCols.ShowDialog() == DialogResult.OK)
{
rowID = mfrmSelectCols.RowIDColumn; //mass tags
var dataCols = mfrmSelectCols.DataColumns;
try
{
var mDTselectedEset1 = ArrangeDataTable(mDTloaded, rowID, dataCols);
mDTselectedEset1.Columns[0].ColumnName = "Row_ID";
mDTselectedEset1 = clsDataTable.RemoveDuplicateRows2(mDTselectedEset1,
mDTselectedEset1.Columns[0].ColumnName);
// handle duplicate rows
mDTselectedEset1.TableName = "Eset";
success = mRConnector.SendTable2RmatrixNumeric("Eset", mDTselectedEset1);
if (mhtDatasets.ContainsKey("Factors"))
{
rcmd = "FactorsValid<-identical(as.array(colnames(factors)),as.array(colnames(Eset)))";
mRConnector.EvaluateNoReturn(rcmd);
FactorsValid = mRConnector.GetSymbolAsBool("FactorsValid");
}
if (!FactorsValid)
{
success = false;
}
if (success)
{
AddDataset2HashTable(mDTselectedEset1);
mRConnector.EvaluateNoReturn("print(dim(Eset))");
mRConnector.EvaluateNoReturn("cat(\"Expressions loaded.\n\")");
}
}
catch (Exception ex)
{
MessageBox.Show("Error: " + ex.Message,
"Exception while extracting data columns");
return false;
}
if (mfrmSelectCols.Proteins && success) // Protein info needs to be loaded ?
{
// loads to mDataTableProtInfo and then sends to R
var mdtProts = LoadProtColumns(mDTloaded, mfrmSelectCols.ProteinIDColumn, rowID);
mdtProts.TableName = "ProtInfo";
AddDataset2HashTable(mdtProts);
}
}
else
{
success = false;
}
#endregion
break;
case enmDataType.PROTINFO:
#region Load Protein Info
var mDTtmp = clsDataTable.LoadFile2DataTableFastCSVReader(mstrLoadedfileName);
if (mDTtmp == null)
{
return false;
}
var mfrmSelectProts = new frmSelectProtInfo
{
PopulateListBox = clsDataTable.DataTableColumns(mDTtmp, false)
};
if (mfrmSelectProts.ShowDialog() == DialogResult.OK)
{
rowID = mfrmSelectProts.RowIDColumn; //mass tags
var protIPI = mfrmSelectProts.ProteinIDColumn;
var mdtProts = LoadProtColumns(mDTtmp, protIPI, rowID);
mdtProts.TableName = "ProtInfo";
AddDataset2HashTable(mdtProts);
}
else
{
success = false;
}
#endregion
break;
case enmDataType.FACTORS:
// Factor files can be CSV files or .txt files
// The first row of the factor definitions file must have a column named Factor, then column names that match the names in the originally loaded Expressions table
// Each subsequent row of the factor file is a new factor name, then the factor value for each dataset
//
// Example factor file (.txt file should use a tab delimiter; CSV file would use commas)
//
// Factor P10A P10B P11A P11B P12A P12B
// Time 0 0 5 5 10 10
// Temperature Hot Cold Hot Cold Hot Cold
#region Load Factors
FactorsValid = true;
var mDTFactors = clsDataTable.LoadFile2DataTableFastCSVReader(mstrLoadedfileName);
if (mDTFactors == null)
{
return false;
}
if (mDTFactors.Rows.Count > frmDefFactors.MAX_LEVELS)
{
MessageBox.Show("Factors file has too many factors; max allowed is " + frmDefFactors.MAX_LEVELS + " factors", "Error", MessageBoxButtons.OK, MessageBoxIcon.Exclamation);
return false;
}
if (mRConnector.SendTable2RmatrixNonNumeric("factors", mDTFactors))
{
try
{
if (mhtDatasets.ContainsKey("Expressions"))
{
rcmd = "FactorsValid<-identical(as.array(colnames(factors)),as.array(colnames(Eset)))";
mRConnector.EvaluateNoReturn(rcmd);
FactorsValid = mRConnector.GetSymbolAsBool("FactorsValid");
}
if (!FactorsValid)
{
success = false;
}
else
{
UpdateFactorInfoArray();
mDTFactors.Columns[0].ColumnName = "Factors";
mDTFactors.TableName = "factors";
mRConnector.EvaluateNoReturn("print(factors)");
mRConnector.EvaluateNoReturn("cat(\"Factors loaded.\n\")");
}
}
catch (Exception ex)
{
MessageBox.Show("Error: " + ex.Message, "Exception while talking to R");
success = false;
}
}
if (success)
{
AddDataset2HashTable(mDTFactors);
}
#endregion
break;
default:
break;
}
return success;
}
/// <summary>
/// Main routine that opens data files (Expressions, Factors)
/// in comma or tab delimited CSVs (and Excel as a future possibility).
/// The main file load routine is in clsDataTable.LoadFile2DataTable(fileName)
/// Duplicates are handled in C#
/// </summary>
private bool OpenFile(string filePath)
{
string rowID;
string rcmd;
var success = true;
bool validFactors;
// check that the file exists before opening it
if (!File.Exists(filePath))
{
return(false);
}
if (!string.Equals(mLoadedFilename, filePath))
{
mLoadedFilename = filePath;
}
var fExt = Path.GetExtension(filePath);
if (string.IsNullOrWhiteSpace(fExt))
{
MessageBox.Show("File does not have an extension (like .csv or .txt); unable to determine file type",
"Error!", MessageBoxButtons.OK, MessageBoxIcon.Exclamation);
return(false);
}
if (!ValidExtension(fExt))
{
MessageBox.Show("File type not allowed (must be csv, tsv, txt, xls, or xlsx)", "Error!");
return(false);
}
var dataLoader = new clsDataTable();
dataLoader.OnError += clsDataTable_OnError;
dataLoader.OnWarning += clsDataTable_OnWarning;
dataLoader.OnProgress += clsDataTable_OnProgress;
switch (mDataSetType)
{
case enmDataType.ESET:
#region Load Expressions
validFactors = true;
mProgressForm.Reset("Loading data");
var esetTable = dataLoader.LoadFile2DataTableFastCSVReader(mLoadedFilename);
if (esetTable == null)
{
string errorMessage;
if (string.IsNullOrWhiteSpace(mProgressForm.ErrorMessage))
{
errorMessage = "Unknown load error";
}
else
{
errorMessage = "Load error: " + mProgressForm.ErrorMessage;
}
MessageBox.Show(errorMessage, "Error", MessageBoxButtons.OK, MessageBoxIcon.Exclamation);
return(false);
}
if (!string.IsNullOrWhiteSpace(mProgressForm.WarningMessage))
{
MessageBox.Show(mProgressForm.WarningMessage, "Warning", MessageBoxButtons.OK, MessageBoxIcon.Information);
}
esetTable.TableName = "AllEset";
// Select columns
var columnSelectionForm = new frmSelectColumns
{
PopulateListBox = clsDataTable.DataTableColumns(esetTable, false),
Proteins = mhtDatasets.ContainsKey("Protein Info")
};
if (columnSelectionForm.ShowDialog() == DialogResult.OK)
{
rowID = columnSelectionForm.RowIDColumn; //mass tags
var dataCols = columnSelectionForm.DataColumns.ToList();
try
{
var filteredDataTable = ArrangeDataTable(esetTable, rowID, dataCols, false);
// Rename the first column from MassTagID (or whatever the user-supplied name is) to Row_ID
filteredDataTable.Columns[0].ColumnName = "Row_ID";
// Remove rows with no data or duplicate data
filteredDataTable = dataLoader.RemoveDuplicateRows2(
filteredDataTable,
filteredDataTable.Columns[0].ColumnName);
// Copy the data into R
filteredDataTable.TableName = "Eset";
success = mRConnector.SendTable2RmatrixNumeric("Eset", filteredDataTable);
if (mhtDatasets.ContainsKey("Factors"))
{
// Simplistic method looking for exact duplicates:
// rcmd = "FactorsValid<-identical(as.array(colnames(factors)),as.array(colnames(Eset)))";
// mRConnector.EvaluateNoReturn(rcmd);
// validFactors = mRConnector.GetSymbolAsBool("FactorsValid");
// Better method for comparing column names, including notifying the user of missing columns
validFactors = ValidateFactors();
}
if (!validFactors)
{
success = false;
}
if (!string.IsNullOrWhiteSpace(mProgressForm.ErrorMessage))
{
MessageBox.Show(mProgressForm.ErrorMessage, "Error", MessageBoxButtons.OK, MessageBoxIcon.Exclamation);
}
if (success)
{
AddDataset2HashTable(filteredDataTable);
mRConnector.EvaluateNoReturn("print(dim(Eset))");
mRConnector.EvaluateNoReturn("cat(\"Expressions loaded.\n\")");
if (!string.IsNullOrWhiteSpace(mProgressForm.WarningMessage))
{
MessageBox.Show(mProgressForm.WarningMessage, "Warning", MessageBoxButtons.OK, MessageBoxIcon.Information);
}
}
}
catch (Exception ex)
{
MessageBox.Show("Error: " + ex.Message,
"Exception while extracting data columns");
return(false);
}
if (columnSelectionForm.Proteins &&
!string.IsNullOrWhiteSpace(columnSelectionForm.ProteinIDColumn) && success)
{
// Load protein info then send to R
var proteinDataTable = LoadProtColumns(esetTable, columnSelectionForm.ProteinIDColumn, rowID,
columnSelectionForm.ProteinMetadataColumns);
proteinDataTable.TableName = "ProtInfo";
AddDataset2HashTable(proteinDataTable);
}
}
else
{
success = false;
}
#endregion
break;
case enmDataType.PROTINFO:
#region Load Protein Info
var proteinInfoTable = dataLoader.LoadFile2DataTableFastCSVReader(mLoadedFilename);
if (proteinInfoTable == null)
{
return(false);
}
var proteinSelectionForm = new frmSelectProtInfo
{
PopulateListBox = clsDataTable.DataTableColumns(proteinInfoTable, false)
};
if (proteinSelectionForm.ShowDialog() == DialogResult.OK)
{
rowID = proteinSelectionForm.RowIDColumn; //mass tags
var proteinIdentifierColumn = proteinSelectionForm.ProteinIDColumn;
var proteinDataTable = LoadProtColumns(proteinInfoTable, proteinIdentifierColumn, rowID,
proteinSelectionForm.ProteinMetadataColumns);
proteinDataTable.TableName = "ProtInfo";
AddDataset2HashTable(proteinDataTable);
}
else
{
success = false;
}
#endregion
break;
case enmDataType.FACTORS:
// Factor files can be CSV files or .txt files
// The first row of the factor definitions file must have a column named Factor, then column names that match the names in the originally loaded Expressions table
// Each subsequent row of the factor file is a new factor name, then the factor value for each dataset
//
// Example factor file (.txt file should use a tab delimiter; CSV file would use commas)
//
// Factor P10A P10B P11A P11B P12A P12B
// Time 0 0 5 5 10 10
// Temperature Hot Cold Hot Cold Hot Cold
#region Load Factors
validFactors = true;
var factorTable = dataLoader.LoadFile2DataTableFastCSVReader(mLoadedFilename);
if (factorTable == null)
{
return(false);
}
if (factorTable.Rows.Count > frmDefFactors.MAX_LEVELS)
{
MessageBox.Show(
"Factors file has too many factors; max allowed is " + frmDefFactors.MAX_LEVELS + " factors",
"Error", MessageBoxButtons.OK, MessageBoxIcon.Exclamation);
return(false);
}
if (factorTable.Columns.Count == 0)
{
MessageBox.Show(
"Factors file is empty; nothing to load",
"Error", MessageBoxButtons.OK, MessageBoxIcon.Exclamation);
return(false);
}
if (mhtDatasets.ContainsKey("Expressions"))
{
// Check for extra sample names in the factorTable and remove them
// The various calls to R that utilize factors do not work properly when extra columns are present
mRConnector.EvaluateNoReturn("esetColNames <- colnames(Eset)");
var esetColNamesFromR = mRConnector.GetSymbolAsStrings("esetColNames");
var esetColNames = new SortedSet <string>();
foreach (var item in esetColNamesFromR.Distinct())
{
esetColNames.Add(item);
}
var extraFactorCols = new List <string>();
var factorColumnName = string.Empty;
foreach (var factorNameToFind in new List <string> {
"Factor", "Factors"
})
{
foreach (DataColumn factorCol in factorTable.Columns)
{
if (string.Equals(factorCol.ColumnName, factorNameToFind, StringComparison.OrdinalIgnoreCase))
{
factorColumnName = factorNameToFind;
break;
}
}
}
if (string.IsNullOrWhiteSpace(factorColumnName))
{
MessageBox.Show(
string.Format(
"The first column of the factors file must be named Factor or Factors; your file has {0}",
factorTable.Columns[0].ColumnName),
"Error", MessageBoxButtons.OK, MessageBoxIcon.Information);
return(false);
}
// Assure that the factor column is the first column
if (factorTable.Columns[factorColumnName].Ordinal > 0)
{
factorTable.Columns[factorColumnName].SetOrdinal(0);
}
foreach (DataColumn factorCol in factorTable.Columns)
{
if (string.Equals(factorCol.ColumnName, factorColumnName, StringComparison.OrdinalIgnoreCase))
{
continue;
}
if (!esetColNames.Contains(factorCol.ColumnName))
{
extraFactorCols.Add(factorCol.ColumnName);
}
}
if (extraFactorCols.Count > 0)
{
foreach (var colToRemove in extraFactorCols)
{
factorTable.Columns.Remove(colToRemove);
}
if (extraFactorCols.Count == 1)
{
MessageBox.Show(
string.Format(
"Removed 1 unknown sample name from the factors file (did not match any expression column names): {0}",
extraFactorCols.First()),
"Warning", MessageBoxButtons.OK, MessageBoxIcon.Information);
}
else
{
MessageBox.Show(
string.Format(
"Removed {0} unknown sample names from the factors file (names did not match expression column names): {1}",
extraFactorCols.Count,
string.Join(", ", extraFactorCols)),
"Warning", MessageBoxButtons.OK, MessageBoxIcon.Information);
}
}
}
if (mRConnector.SendTable2RmatrixNonNumeric("factors", factorTable))
{
try
{
if (mhtDatasets.ContainsKey("Expressions"))
{
validFactors = ValidateFactors();
}
if (!validFactors)
{
success = false;
}
else
{
UpdateFactorInfoArray();
factorTable.Columns[0].ColumnName = "Factors";
factorTable.TableName = "factors";
mRConnector.EvaluateNoReturn("print(factors)");
mRConnector.EvaluateNoReturn("cat(\"Factors loaded.\n\")");
}
}
catch (Exception ex)
{
MessageBox.Show("Error: " + ex.Message, "Exception while talking to R");
success = false;
}
}
if (success)
{
AddDataset2HashTable(factorTable);
}
#endregion
break;
}
return(success);
}