public TCGADatatableBuilderUI()
{
InitializeComponent();
rootDir.SetDirectoryArgument("TCGARoot", "TCGA Data Root");
targetFile.FileArgument = new SaveFileArgument("Target Data", "tsv");
lbDataTypes.Items.AddRange(TCGATechnology.GetTechnologyNames().ToArray());
lbSampleTypes.Items.AddRange(TCGASampleCode.GetSampleCodes().OrderBy(m => m.Code).ToList().ConvertAll(m => string.Format("{0}, {1}", m.ShortLetterCode, m.Definition)).ToArray());
this.Text = Constants.GetSQHTitle(title, version);
}
public IList <TCGASampleCode> GetTCGASampleCodes()
{
List <TCGASampleCode> result = new List <TCGASampleCode>();
foreach (var s in TCGASampleCodeStrings)
{
var code = TCGASampleCode.Find(s);
if (code == null)
{
throw new ArgumentException("Cannot find sample code for {0}", s);
}
result.Add(code);
}
return(result);
}
public override bool PrepareOptions()
{
if (!Directory.Exists(this.TCGADirectory))
{
ParsingErrors.Add(string.Format("Directory not exists {0}.", this.TCGADirectory));
return(false);
}
try
{
TCGATechnology.Parse(this.DataType);
}
catch (Exception ex)
{
ParsingErrors.Add(ex.Message);
return(false);
}
foreach (var tumor in this.TumorTypes)
{
var tumordir = this.TCGADirectory + "/" + tumor;
if (!Directory.Exists(tumordir))
{
ParsingErrors.Add(string.Format("Directory not exists {0}.", tumordir));
return(false);
}
}
if (TCGASampleCodeStrings == null || TCGASampleCodeStrings.Count == 0)
{
TCGASampleCodeStrings = (from v in TCGASampleCode.GetSampleCodes()
select v.ShortLetterCode).ToList();
}
else
{
try
{
GetTCGASampleCodes();
}
catch (Exception ex)
{
ParsingErrors.Add(ex.Message);
return(false);
}
}
try
{
GetTechnology();
}
catch (Exception ex)
{
ParsingErrors.Add(ex.Message);
return(false);
}
if (this.Platforms == null || this.Platforms.Count == 0)
{
var tec = GetTechnology();
this.Platforms = (from tumor in TumorTypes
let dir = Path.Combine(this.TCGADirectory, tumor)
let tecdir = tec.GetTechnologyDirectory(dir)
from subdir in Directory.GetDirectories(tecdir)
select Path.GetFileName(subdir)).Distinct().OrderBy(m => m).ToList();
}
return(true);
}
public override IEnumerable <string> Process()
{
var result = new List <string>();
Dictionary <string, BarInfo> barMap;
barMap = new Dictionary <string, BarInfo>();
foreach (var tumor in _options.TumorTypes)
{
var curMap = TCGAUtils.GetBarcodeFileMap(_options.TCGADirectory,
_options.GetTechnology(), tumor, _options.Platforms, _options.GetTCGASampleCodes().ToArray());
foreach (var v in curMap)
{
barMap[GetSampleKey(tumor, v.Key)] = v.Value;
}
}
var headers = new List <string>();
var clindata = new Dictionary <string, IAnnotation>();
foreach (var tumor in _options.TumorTypes)
{
ReadClinData(clindata, tumor, headers);
}
Console.WriteLine("{0} patient clinical information readed", clindata.Count);
List <string> noclinical = new List <string>();
var keyvalues = barMap.ToList();
foreach (var bm in keyvalues)
{
if (!clindata.ContainsKey(GetSampleKey(GetTumorType(bm.Key), bm.Value.Paticipant)))
{
noclinical.Add(bm.Key);
Console.Error.WriteLine(string.Format("Cannot find clinical data for patient {0}", bm.Value.Paticipant));
if (_options.WithClinicalInformationOnly)
{
barMap.Remove(bm.Key);
}
}
}
Progress.SetMessage("Reading data ...");
Func <double, double> getValue;
var valueMap = GetData(barMap, out getValue);
var genes = GetCommonGenes(valueMap);
var samples = valueMap.Keys.OrderBy(m => m).ToList();
Progress.SetMessage("Saving data ...");
result.Add(_options.OutputFile);
result.Add(_options.DesignFile);
if (_options.TumorTypes.Count > 1)
{
using (var sw = new StreamWriter(_options.OutputFile))
{
sw.WriteLine("Gene\t{0}", samples.Merge("\t"));
foreach (var gene in genes)
{
sw.Write(gene);
foreach (var sample in samples)
{
sw.Write("\t{0}", getValue(valueMap[sample][gene]));
}
sw.WriteLine();
}
}
using (var sw = new StreamWriter(_options.DesignFile))
{
sw.Write("Sample\tBarcode\tPatient\tTumorType\tPlatform\tSampleType\tSampleTypeDescription");
if (headers.Count > 0)
{
sw.WriteLine("\t{0}", headers.Merge("\t"));
}
else
{
sw.WriteLine();
}
foreach (var entry in barMap)
{
var tumor = GetTumorType(entry.Key);
var type = TCGASampleCode.Find(entry.Value.Sample);
sw.Write("{0}\t{1}\t{2}\t{3}\t{4}\t{5}\t{6}", entry.Key, entry.Value.BarCode, entry.Value.Paticipant, tumor, entry.Value.Platform, type.ShortLetterCode,
type.Definition);
var key = GetSampleKey(tumor, entry.Value.Paticipant);
var vdata = clindata.ContainsKey(key) ? clindata[key] : new Annotation();
foreach (var header in headers)
{
if (vdata.Annotations.ContainsKey(header))
{
sw.Write("\t{0}", vdata.Annotations[header]);
}
else
{
sw.Write("\t");
}
}
sw.WriteLine();
}
}
}
else
{
using (var sw = new StreamWriter(_options.OutputFile))
{
sw.WriteLine("Gene\t{0}", (from s in samples select s.StringAfter("_")).Merge("\t"));
foreach (var gene in genes)
{
sw.Write(gene);
foreach (var sample in samples)
{
sw.Write("\t{0}", getValue(valueMap[sample][gene]));
}
sw.WriteLine();
}
}
using (var sw = new StreamWriter(_options.DesignFile))
{
sw.Write("Sample\tBarcode\tPatient\tTumorType\tPlatform\tSampleType\tSampleTypeDescription");
if (headers.Count > 0)
{
sw.WriteLine("\t{0}", headers.Merge("\t"));
}
else
{
sw.WriteLine();
}
foreach (var entry in barMap)
{
var tumor = _options.TumorTypes.First();
var type = TCGASampleCode.Find(entry.Value.Sample);
sw.Write("{0}\t{1}\t{2}\t{3}\t{4}\t{5}\t{6}", entry.Key.StringAfter("_"), entry.Value.BarCode, entry.Value.Paticipant, tumor, entry.Value.Platform, type.ShortLetterCode,
type.Definition);
var key = GetSampleKey(tumor, entry.Value.Paticipant);
var vdata = clindata.ContainsKey(key) ? clindata[key] : new Annotation();
foreach (var header in headers)
{
if (vdata.Annotations.ContainsKey(header))
{
sw.Write("\t{0}", vdata.Annotations[header]);
}
else
{
sw.Write("\t");
}
}
sw.WriteLine();
}
}
var clinicalOptions = new TCGAClinicalInformationBuilderOptions()
{
ClinicalFile = TCGAUtils.GetClinicPatientFile(_options.TCGADirectory, _options.TumorTypes.First()),
DataFile = _options.OutputFile,
ThrowException = false,
};
result.AddRange(new TCGAClinicalInformationBuilder(clinicalOptions)
{
Progress = this.Progress
}.Process());
}
Progress.End();
if (noclinical.Count == 0)
{
return(result.ToArray());
}
else
{
return(new[] { string.Format("There are {0} samples without patient information:\n {1}\n\nResult have been saved to:\n {2}", noclinical.Count, noclinical.Merge("\n "), result.Merge("\n ")) });
}
}