public Dictionary <string, SampleItem> ParseDataset(string datasetDirectory)
{
var result = new Dictionary <string, SampleItem>();
var columns = (from m in maps
where !string.IsNullOrEmpty(m.PropertyName)
select m).ToDictionary(m => m.AnnotationName);
var dvs = (from m in maps.DefaultValues
where !string.IsNullOrEmpty(m.Value)
select m).ToList();
var files = GeoUtils.GetGsmNameFileMap(datasetDirectory);
var datasetName = Path.GetFileName(datasetDirectory);
var dataMap = new RawSampleInfoReader().ReadDescriptionFromDirectory(datasetDirectory);
foreach (var sampleName in dataMap.Keys)
{
string filename;
if (files.TryGetValue(sampleName.ToUpper(), out filename))
{
var sample = new SampleItem()
{
Dataset = datasetName,
Sample = sampleName,
SampleFile = filename
};
result[sampleName] = sample;
var qsMap = dataMap[sampleName];
foreach (var column in columns)
{
List <string> values;
if (qsMap.TryGetValue(column.Key, out values))
{
sample.Annotations[column.Value.PropertyName] = values.FirstOrDefault();
}
}
foreach (var dv in dvs)
{
if (!sample.Annotations.ContainsKey(dv.PropertyName))
{
sample.Annotations[dv.PropertyName] = dv.Value;
}
}
}
}
return(result);
}
public override IEnumerable <string> Process()
{
var sdata = new RawSampleInfoReader().ReadDescriptionFromDirectory(options.InputDirectory);
var data = sdata.ToList().ToDictionary(m => m.Key.ToUpper(), m => m.Value);
var files = GeoUtils.GetGsmNameFileMap(options.InputDirectory);
var samples = (from k in files.Keys select k.ToUpper()).OrderBy(m => m).ToList();
var columns = (from d in data.Values
from col in d.Keys
select col).Distinct().OrderBy(m => m).ToList();
bool bError = false;
var errorFile = options.OutputFile + ".error";
using (var sw = new StreamWriter(options.OutputFile))
using (var swErr = new StreamWriter(errorFile))
{
sw.WriteLine("Sample\t{0}", columns.Merge("\t"));
foreach (var sample in samples)
{
if (!data.ContainsKey(sample))
{
var error = string.Format("Cannot find {0} in {1}", sample, options.InputDirectory);
swErr.WriteLine(error);
Progress.SetMessage(error);
bError = true;
continue;
}
var dic = data[sample];
sw.Write(sample);
foreach (var column in columns)
{
if (dic.ContainsKey(column))
{
sw.Write("\t{0}", dic[column].Merge(" ! "));
}
else
{
sw.Write("\t");
}
}
sw.WriteLine();
}
}
if (!bError)
{
File.Delete(errorFile);
}
return(new string[] { options.OutputFile });
}
public Dictionary<string, SampleItem> ParseDataset(string datasetDirectory)
{
var result = new Dictionary<string, SampleItem>();
var columns = (from m in maps
where !string.IsNullOrEmpty(m.PropertyName)
select m).ToDictionary(m => m.AnnotationName);
var dvs = (from m in maps.DefaultValues
where !string.IsNullOrEmpty(m.Value)
select m).ToList();
var files = GeoUtils.GetGsmNameFileMap(datasetDirectory);
var datasetName = Path.GetFileName(datasetDirectory);
var dataMap = new RawSampleInfoReader().ReadDescriptionFromDirectory(datasetDirectory);
foreach (var sampleName in dataMap.Keys)
{
string filename;
if (files.TryGetValue(sampleName.ToUpper(), out filename))
{
var sample = new SampleItem()
{
Dataset = datasetName,
Sample = sampleName,
SampleFile = filename
};
result[sampleName] = sample;
var qsMap = dataMap[sampleName];
foreach (var column in columns)
{
List<string> values;
if (qsMap.TryGetValue(column.Key, out values))
{
sample.Annotations[column.Value.PropertyName] = values.FirstOrDefault();
}
}
foreach (var dv in dvs)
{
if (!sample.Annotations.ContainsKey(dv.PropertyName))
{
sample.Annotations[dv.PropertyName] = dv.Value;
}
}
}
}
return result;
}
public override IEnumerable<string> Process()
{
var sdata = new RawSampleInfoReader().ReadDescriptionFromDirectory(options.InputDirectory);
var data = sdata.ToList().ToDictionary(m => m.Key.ToUpper(), m => m.Value);
var files = GeoUtils.GetGsmNameFileMap(options.InputDirectory);
var samples = (from k in files.Keys select k.ToUpper()).OrderBy(m => m).ToList();
var columns = (from d in data.Values
from col in d.Keys
select col).Distinct().OrderBy(m => m).ToList();
bool bError = false;
var errorFile = options.OutputFile + ".error";
using (var sw = new StreamWriter(options.OutputFile))
using (var swErr = new StreamWriter(errorFile))
{
sw.WriteLine("Sample\t{0}", columns.Merge("\t"));
foreach (var sample in samples)
{
if (!data.ContainsKey(sample)){
var error = string.Format("Cannot find {0} in {1}", sample, options.InputDirectory);
swErr.WriteLine(error);
Progress.SetMessage(error);
bError = true;
continue;
}
var dic = data[sample];
sw.Write(sample);
foreach (var column in columns)
{
if (dic.ContainsKey(column))
{
sw.Write("\t{0}", dic[column].Merge(" ! "));
}
else
{
sw.Write("\t");
}
}
sw.WriteLine();
}
}
if (!bError)
{
File.Delete(errorFile);
}
return new string[] { options.OutputFile };
}
public void ParseDataset(string datasetDirectory, Dictionary<string, BreastCancerSampleItem> sampleMap)
{
var files = GeoUtils.GetGsmNameFileMap(datasetDirectory);
var dirname = Path.GetFileName(datasetDirectory);
var map = new RawSampleInfoReader().ReadDescriptionFromDirectory(datasetDirectory);
foreach (var key in map.Keys)
{
if (files.ContainsKey(key.ToLower()))
{
if (!sampleMap.ContainsKey(key))
{
sampleMap[key] = new BreastCancerSampleItem(dirname, key);
}
var sample = sampleMap[key];
var qsMap = map[key];
//parse information
foreach (var question in qsMap.Keys)
{
if (converters.ContainsKey(question))
{
var converter = converters[question];
var answer = qsMap[question].First();
converter.SetProperty(sample, answer);
}
}
//set defaultvalue
foreach (var dfConverter in defaultConverters)
{
dfConverter.Key.SetProperty(sample, dfConverter.Value);
}
}
}
}
public void NewFromData(string subdir)
{
var siformat = Directory.GetFiles(subdir, "*.siformat");
TextFileDefinition prefile = new TextFileDefinition();
if (siformat.Length > 0)
{
prefile.ReadFromFile(siformat[0]);
bool bFound = false;
prefile.ForEach(m =>
{
if (m.PropertyName.Equals("TumorStage"))
{
m.PropertyName = "TumorStatus";
bFound = true;
}
if (m.PropertyName.Equals("Metastasis"))
{
m.PropertyName = "MetastasisStatus";
bFound = true;
}
});
if (bFound)
{
prefile.WriteToFile(siformat[0]);
}
}
var map = new RawSampleInfoReader().ReadDescriptionFromDirectory(subdir);
lastDirectory = subdir;
lastFile = String.Empty;
var files = new HashSet<string>(from f in CelFile.GetCelFiles(subdir, false)
select GeoUtils.GetGsmName(f));
Dictionary<string, HashSet<string>> headers = new Dictionary<string, HashSet<string>>();
foreach (var m in map)
{
var gsm = m.Key.ToLower();
if (!files.Contains(gsm))
{
continue;
}
var curmap = m.Value;
foreach (var entry in curmap)
{
if (!headers.ContainsKey(entry.Key))
{
headers[entry.Key] = new HashSet<string>();
}
headers[entry.Key].UnionWith(entry.Value);
}
}
ClearDataSource();
items.Clear();
foreach (var part in headers)
{
items.Add(new FileDefinitionItem()
{
AnnotationName = part.Key,
Example = (from v in part.Value
orderby v
select v).Merge(";")
});
}
foreach (var olditem in prefile)
{
if (!string.IsNullOrEmpty(olditem.PropertyName))
{
var newitem = items.Find(m => m.AnnotationName.Equals(olditem.AnnotationName));
if (newitem != null)
{
newitem.PropertyName = olditem.PropertyName;
}
}
}
items.DefaultValues.Clear();
foreach (var olddv in prefile.DefaultValues)
{
if (propertyNames.Contains(olddv.PropertyName))
{
items.DefaultValues.Add(new DefaultValue()
{
PropertyName = olddv.PropertyName,
Value = olddv.Value
});
}
}
items.Sort((m1, m2) => m1.AnnotationName.CompareTo(m2.AnnotationName));
UpdateDataSource();
this.Text = title + " - " + Path.GetFileName(subdir);
}
private void btnTest_Click(object sender, EventArgs e)
{
if (dlgOpenDirectory.ShowDialog() == System.Windows.Forms.DialogResult.OK)
{
FormToDefinition();
var parser = new PropertyMappingParser(items);
var map = new Dictionary<string, BreastCancerSampleItem>();
parser.ParseDataset(dlgOpenDirectory.SelectedPath, map);
var lst = (from v in map.Values
orderby v.Sample
select v).ToList();
var form = new BreastCancerSampleInformationForm();
var reader = new RawSampleInfoReader();
form.SetRawInfoReader(reader, Path.GetFileNameWithoutExtension(dlgOpenDirectory.SelectedPath));
form.SetDataSource(lst);
form.ShowDialog();
}
}
public void NewFromData(string subdir)
{
try
{
var siformat = Directory.GetFiles(subdir, "*.siformat");
TextFileDefinition prefile = new TextFileDefinition();
if (siformat.Length > 0)
{
prefile.ReadFromFile(siformat[0]);
}
var map = new RawSampleInfoReader().ReadDescriptionFromDirectory(subdir);
LastDirectory = subdir;
lastFile = String.Empty;
var files = new HashSet<string>(from f in CelFile.GetCelFiles(subdir, false)
select GeoUtils.GetGsmName(f));
Dictionary<string, HashSet<string>> headers = new Dictionary<string, HashSet<string>>();
foreach (var m in map)
{
var gsm = m.Key.ToUpper();
if (!files.Contains(gsm))
{
continue;
}
var curmap = m.Value;
foreach (var entry in curmap)
{
if (!headers.ContainsKey(entry.Key))
{
headers[entry.Key] = new HashSet<string>();
}
headers[entry.Key].UnionWith(entry.Value);
}
}
ClearDataSource();
items.Clear();
foreach (var part in headers)
{
items.Add(new FileDefinitionItem()
{
AnnotationName = part.Key,
Example = (from v in part.Value
orderby v
select v).Merge(";")
});
}
foreach (var olditem in prefile)
{
if (!string.IsNullOrEmpty(olditem.PropertyName))
{
var newitem = items.Find(m => m.AnnotationName.Equals(olditem.AnnotationName));
if (newitem != null)
{
newitem.PropertyName = olditem.PropertyName;
}
}
}
items.DefaultValues.Clear();
items.Sort((m1, m2) => m1.AnnotationName.CompareTo(m2.AnnotationName));
UpdateDataSource();
label1.Text = "Annotation/property mapping - " + Path.GetFileName(subdir);
dlgOpenDirectory.SelectedPath = subdir;
dlgSaveFormatFile.FileName = Path.Combine(subdir, Path.GetFileName(subdir) + ".siformat");
}
catch (Exception ex)
{
MessageBox.Show(this, ex.Message, "Error", MessageBoxButtons.OK, MessageBoxIcon.Error);
}
}
public void NewFromData(string subdir)
{
try
{
var siformat = Directory.GetFiles(subdir, "*.siformat");
TextFileDefinition prefile = new TextFileDefinition();
if (siformat.Length > 0)
{
prefile.ReadFromFile(siformat[0]);
}
var map = new RawSampleInfoReader().ReadDescriptionFromDirectory(subdir);
LastDirectory = subdir;
lastFile = String.Empty;
var files = new HashSet <string>(from f in CelFile.GetCelFiles(subdir, false)
select GeoUtils.GetGsmName(f));
Dictionary <string, HashSet <string> > headers = new Dictionary <string, HashSet <string> >();
foreach (var m in map)
{
var gsm = m.Key.ToUpper();
if (!files.Contains(gsm))
{
continue;
}
var curmap = m.Value;
foreach (var entry in curmap)
{
if (!headers.ContainsKey(entry.Key))
{
headers[entry.Key] = new HashSet <string>();
}
headers[entry.Key].UnionWith(entry.Value);
}
}
ClearDataSource();
items.Clear();
foreach (var part in headers)
{
items.Add(new FileDefinitionItem()
{
AnnotationName = part.Key,
Example = (from v in part.Value
orderby v
select v).Merge(";")
});
}
foreach (var olditem in prefile)
{
if (!string.IsNullOrEmpty(olditem.PropertyName))
{
var newitem = items.Find(m => m.AnnotationName.Equals(olditem.AnnotationName));
if (newitem != null)
{
newitem.PropertyName = olditem.PropertyName;
}
}
}
items.DefaultValues.Clear();
items.Sort((m1, m2) => m1.AnnotationName.CompareTo(m2.AnnotationName));
UpdateDataSource();
label1.Text = "Annotation/property mapping - " + Path.GetFileName(subdir);
dlgOpenDirectory.SelectedPath = subdir;
dlgSaveFormatFile.FileName = Path.Combine(subdir, Path.GetFileName(subdir) + ".siformat");
}
catch (Exception ex)
{
MessageBox.Show(this, ex.Message, "Error", MessageBoxButtons.OK, MessageBoxIcon.Error);
}
}