void TestFileGenerator::WriteSmallCelFileNoStdev()
{
CelFileData data("small_cel_file_no_stdev");
data.SetIntensityCount(25);
data.SetStdDevCount(0);
data.SetPixelCount(25);
data.SetOutlierCount(2);
data.SetMaskCount(3);
data.SetRows(5);
data.SetCols(5);
data.SetArrayType(L"Hg-small");
data.SetAlgorithmName(L"Feature Extraction");
ParameterNameValueType nvt;
nvt.SetName(L"percentile");
nvt.SetValueFloat(0.75f);
data.AddAlgorithmParameter(nvt);
nvt.SetName(L"outlierlow");
nvt.SetValueFloat(1.004f);
data.AddAlgorithmParameter(nvt);
CelFileWriter* writer = new CelFileWriter(data);
FloatVector vInten;
Int16Vector vPixels;
for (int i=0; i<25; ++i)
{
vInten.push_back(100.0f*i);
vPixels.push_back(25);
}
// Do some writing
writer->WriteIntensities(vInten);
writer->WritePixels(vPixels);
// XYCoordVector
XYCoordVector outlier;
XYCoord xy(0,0);
outlier.push_back(xy);
xy.xCoord = 1;
xy.yCoord = 2;
outlier.push_back(xy);
writer->WriteOutlierCoords(outlier);
XYCoordVector masked;
xy.xCoord = 1;
xy.yCoord = 0;
masked.push_back(xy);
xy.xCoord = 2;
xy.yCoord = 1;
masked.push_back(xy);
xy.xCoord = 3;
xy.yCoord = 2;
masked.push_back(xy);
writer->WriteMaskCoords(masked);
delete writer;
}
void CDFData::SetUInt32ToGenericHdr(const std::wstring& name, u_int32_t value)
{
ParameterNameValueType paramType;
paramType.SetName(name);
paramType.SetValueUInt32(value);
GenericDataHeader* hdr = genericData.Header().GetGenericDataHdr();
hdr->AddNameValParam(paramType);
}
void CHPQuantificationData::SetWStringToGenericHdr(const std::wstring& name, const std::wstring value, int32_t reserve)
{
ParameterNameValueType paramType;
paramType.SetName(name);
paramType.SetValueText(value, reserve);
GenericDataHeader* hdr = genericData.Header().GetGenericDataHdr();
hdr->AddNameValParam(paramType);
}
void CHPData::SetFloatToGenericHdr(const std::wstring& name, float value)
{
ParameterNameValueType paramType;
paramType.SetName(name);
paramType.SetValueFloat(value);
GenericDataHeader* hdr = genericData.Header().GetGenericDataHdr();
hdr->AddNameValParam(paramType);
}
void CDFData::SetRefSequence(const std::string &seq)
{
GenericDataHeader* gPtr = genericData.Header().GetGenericDataHdr();
ParameterNameValueType paramType;
paramType.SetName(CDF_REFSEQ_PARAM);
paramType.SetValueAscii(seq);
gPtr->AddNameValParam(paramType);
}
void TestFileGenerator::WriteOutGenericDATDataFileNoGrid()
{
GenericDataHeader gdh;
AddStandardGenericDataHeader(gdh);
// Fill the DataGroupHeader
DataGroupHeader dch;
dch.SetName(L"First Data Cube");
// Fill the DataSetHeader
DataSetHeader dph;
dph.SetName(L"acquired data");
ParameterNameValueType nvt;
nvt.SetName(L"Scanner");
nvt.SetValueText(L"M10");
dph.AddNameValParam(nvt);
nvt.SetName(L"Pixel Size");
nvt.SetValueFloat(0.051f);
dph.AddNameValParam(nvt);
dph.AddColumn(UShortColumn(L"Pixel"));
int32_t rows = 100;
dph.SetRowCnt(rows);
dch.AddDataSetHdr(dph);
// Set the FileHeader
FileHeader fh;
fh.SetFilename("test.file.data_dat");
fh.SetGenericDataHdr(gdh);
fh.AddDataGroupHdr(dch);
// Create the generic file writer
GenericFileWriter gfWriter(&fh);
gfWriter.WriteHeader();
DataGroupWriterIt dcwBegin, dcwEnd;
gfWriter.GetDataGroupWriters(dcwBegin, dcwEnd);
DataGroupWriter d = *dcwBegin;
dcwBegin->WriteHeader();
DataSetWriterIt dpwBegin, dpwEnd;
dcwBegin->GetDataSetWriters(dpwBegin, dpwEnd);
dpwBegin->WriteHeader();
// Write out the data
for( int32_t i=0; i < rows; ++i )
{
u_int16_t value = (u_int16_t)(i*10+i);
dpwBegin->Write(value);
}
dpwBegin->UpdateNextDataSetOffset();
dcwBegin->Close();
}
std::wstring CalvinCELDataAdapter::GetAlgorithmParameter(const wchar_t *tag)
{
std::wstring result;
ParameterNameValueType param;
if (calvinCel.FindAlgorithmParameter(tag, param))
{
result = param.ToString();
}
return result;
}
ParameterNameValueType CHPData::GetAlgParam(const std::wstring& tag)
{
std::wstring name = ALGORITHM_PARAM_NAME_PREFIX_S + tag;
ParameterNameValueType paramType;
GenericDataHeader* hdr = genericData.Header().GetGenericDataHdr();
hdr->FindNameValParam(name, paramType);
ParameterNameValueType type = paramType;
type.SetName(tag);
return type;
}
ParameterNameValueType CHPData::GetChipSum(const std::wstring& tag)
{
std::wstring name = CHIP_SUMMARY_PARAMETER_NAME_PREFIX_S + tag;
ParameterNameValueType paramType;
GenericDataHeader* hdr = genericData.Header().GetGenericDataHdr();
hdr->FindNameValParam(name, paramType);
ParameterNameValueType type = paramType;
type.SetName(tag);
return type;
}
void CHPMultiDataDataTest::test_SumParams()
{
ParameterNameValueTypeList params;
ParameterNameValueType param;
CHPMultiDataData data;
param.SetName(L"n1");
param.SetValueText(L"v1");
params.push_back(param);
param.SetName(L"n2");
param.SetValueText(L"v2");
params.push_back(param);
data.AddSummaryParams(params);
ParameterNameValueTypeList params_out = data.GetSummaryParams();
CPPUNIT_ASSERT(params_out.size() == 2);
ParameterNameValueTypeList::iterator it = params_out.begin();
param = *it;
CPPUNIT_ASSERT(param.GetName() == L"n1");
CPPUNIT_ASSERT(param.GetValueText() == L"v1");
++it;
param = *it;
CPPUNIT_ASSERT(param.GetName() == L"n2");
CPPUNIT_ASSERT(param.GetValueText() == L"v2");
}
void CHPQuantificationDataTest::test_AlgParams()
{
ParameterNameValueTypeList params;
ParameterNameValueType param;
CHPQuantificationData data;
param.SetName(L"n1");
param.SetValueText(L"v1");
params.push_back(param);
param.SetName(L"n2");
param.SetValueText(L"v2");
params.push_back(param);
data.AddAlgParams(params);
ParameterNameValueTypeList params_out = data.GetAlgParams();
CPPUNIT_ASSERT(params_out.size() == 2);
ParameterNameValueTypeList::iterator it = params_out.begin();
param = *it;
CPPUNIT_ASSERT(param.GetName() == L"n1");
CPPUNIT_ASSERT(param.GetValueText() == L"v1");
++it;
param = *it;
CPPUNIT_ASSERT(param.GetName() == L"n2");
CPPUNIT_ASSERT(param.GetValueText() == L"v2");
}
void CHPMultiDataData::AddAlgParams(const ParameterNameValueTypeList& params)
{
ParameterNameValueType param;
GenericDataHeader* hdr = genericData.Header().GetGenericDataHdr();
for (ParameterNameValueTypeList::const_iterator it=params.begin(); it != params.end(); it++)
{
param = *it;
param.SetName(ALGORITHM_PARAM_NAME_PREFIX_S + param.GetName());
hdr->AddNameValParam(param);
}
}
void CHPQuantificationData::AddSummaryParams(const ParameterNameValueTypeList& params)
{
ParameterNameValueType param;
GenericDataHeader* hdr = genericData.Header().GetGenericDataHdr();
for (ParameterNameValueTypeList::const_iterator it=params.begin(); it != params.end(); it++)
{
param = *it;
param.SetName(CHIP_SUMMARY_PARAMETER_NAME_PREFIX_S + param.GetName());
hdr->AddNameValParam(param);
}
}
int32_t CHPData::GetInt32FromGenericHdr(const std::wstring& name)
{
int32_t result = 0;
ParameterNameValueType paramType;
GenericDataHeader* hdr = genericData.Header().GetGenericDataHdr();
if (hdr->FindNameValParam(name, paramType))
{
result = paramType.GetValueInt32();
}
return result;
}
std::wstring CHPQuantificationData::GetWStringFromGenericHdr(const std::wstring& name)
{
std::wstring result;
ParameterNameValueType paramType;
GenericDataHeader* hdr = genericData.Header().GetGenericDataHdr();
if (hdr->FindNameValParam(name, paramType))
{
result = paramType.GetValueText();
}
return result;
}
std::string CDFData::GetRefSequence()
{
std::string result;
ParameterNameValueType paramType;
GenericDataHeader* hdr = genericData.Header().GetGenericDataHdr();
if (hdr->FindNameValParam(CDF_REFSEQ_PARAM, paramType))
{
result = paramType.GetValueAscii();
}
return result;
}
void DATFileUpdaterTest::AddParameters(DATData& data)
{
ParameterNameValueType nvt;
nvt.SetName(L"Santa Clara");
nvt.SetValueUInt8(56);
data.AddGridAlignmentAlgorithmParameter(nvt);
nvt.SetName(L"San Mateo");
nvt.SetValueFloat(4.56f);
data.AddGridAlignmentAlgorithmParameter(nvt);
}
void CHPMultiDataData::AddAppMetaInfo(const ParameterNameValueTypeList& params)
{
ParameterNameValueType param;
GenericDataHeader* hdr = genericData.Header().GetGenericDataHdr();
for (ParameterNameValueTypeList::const_iterator it=params.begin(); it != params.end(); ++it)
{
param = *it;
param.SetName(APPLICATION_META_INFO_PREFIX_S + param.GetName());
hdr->AddNameValParam(param);
}
}
void CELConversionUtilities::FindAndAddAlgorithmParameter(string name, string defaultValue, CelFileData &inFile, CCELFileData &outFile)
{
ParameterNameValueType nvt;
if (inFile.FindAlgorithmParameter(StringUtils::ConvertMBSToWCS(name), nvt))
{
outFile.AddAlgorithmParameter(name.c_str(), StringUtils::ConvertWCSToMBS(nvt.ToString()).c_str());
}
else
{
outFile.AddAlgorithmParameter(name.c_str(), defaultValue.c_str());
}
}
bool DataSetHeader::FindNameValParam(const std::wstring& name, ParameterNameValueType& result) const
{
ParameterNameValueType t;
t.SetName(name);
t.SetValueText(L"");
ParameterNameValueTypeConstIt found = FindNameValParam(t);
if (found != nameValParams.end())
{
result = *found;
return true;
}
return false;
}
void CDFDataTest::GetGenericDataTest()
{
// Put something into the header.
data->SetArrayCols(33);
// Get the generic data.
GenericData& gd = data->GetGenericData();
// Try to read the thing from the header.
ParameterNameValueType pvt;
CPPUNIT_ASSERT(gd.Header().GetGenericDataHdr()->FindNameValParam(CDF_COLS_PARAM, pvt));
CPPUNIT_ASSERT(pvt.GetValueUInt32() == 33);
}
void CDFDataTest::FileHeaderTest()
{
// Put something into the header.
data->SetArrayCols(45);
// Get the header.
FileHeader* fh = data->GetFileHeader();
CPPUNIT_ASSERT(fh);
// Try to read the thing from the header.
ParameterNameValueType pvt;
CPPUNIT_ASSERT(fh->GetGenericDataHdr()->FindNameValParam(CDF_COLS_PARAM, pvt));
CPPUNIT_ASSERT(pvt.GetValueUInt32() == 45);
}
void DataGroupHeaderTest::FindDataSetHeaderTest()
{
// Create DataSetHeaders
DataSetHeader dph1;
dph1.SetName(L"pixel intensity");
ParameterNameValueType param;
param.SetName(L"Scanner");
param.SetValueText(L"M10");
dph1.AddNameValParam(param);
dph1.AddUShortColumn(L"");
dph1.SetRowCnt(1);
DataSetHeader dph2;
dph2.SetName(L"grid coordinates");
param.SetName(L"Corner Pattern");
param.SetValueText(L"Checkerboard");
dph2.AddNameValParam(param);
dph2.AddUShortColumn(L"");
dph2.AddUShortColumn(L"");
dph2.AddUShortColumn(L"");
dph2.AddUShortColumn(L"");
dph2.AddUShortColumn(L"");
dph2.AddUShortColumn(L"");
dph2.AddUShortColumn(L"");
dph2.AddUShortColumn(L"");
dph2.AddUShortColumn(L"");
dph2.AddUShortColumn(L"");
dph2.AddUShortColumn(L"");
dph2.AddUShortColumn(L"");
dph2.AddUShortColumn(L"");
dph2.AddUShortColumn(L"");
dph2.AddUShortColumn(L"");
dph2.AddUShortColumn(L"");
dph2.SetRowCnt(1);
header->AddDataSetHdr(dph1);
header->AddDataSetHdr(dph2);
const DataSetHeader* dph = header->FindDataSetHeader(L"none");
CPPUNIT_ASSERT(0 == dph);
dph = header->FindDataSetHeader(dph1.GetName());
CPPUNIT_ASSERT(0 != dph);
CPPUNIT_ASSERT(dph->GetName() == dph1.GetName());
dph = header->FindDataSetHeader(dph2.GetName());
CPPUNIT_ASSERT(0 != dph);
CPPUNIT_ASSERT(dph->GetName() == dph2.GetName());
}
/*
* Get the DAT header from the DAT file header object.
*/
string CELConversionUtilities::GetDatHeader(CelFileData &inFile)
{
std::wstring datHeader;
GenDataHdrVectorIt begin, end;
inFile.GetFileHeader()->GetGenericDataHdr()->GetParentIterators(begin, end);
// Find the DAT generic header
for (GenDataHdrVectorIt ii = begin; ii != end; ++ii)
{
std::string s = ii->GetFileTypeId();
if (ii->GetFileTypeId() == SCAN_ACQUISITION_DATA_TYPE)
{
// found the right header, now look for the parameter
ParameterNameValueType nvt;
if (ii->FindNameValParam(DAT_HEADER_PARAM_NAME, nvt))
{
if (nvt.GetParameterType() == ParameterNameValueType::TextType)
datHeader = nvt.GetValueText();
}
else if (ii->FindNameValParam(PARTIAL_DAT_HEADER_PARAM_NAME, nvt))
{
if (nvt.GetParameterType() == ParameterNameValueType::TextType)
{
std::wstring partialDatHeader = nvt.GetValueText();
u_int16_t min = 0;
u_int16_t max = 0;
// Find the max and min parameters and append to the string.
if (ii->FindNameValParam(MAX_PIXEL_INTENSITY_PARAM_NAME, nvt))
{
if (nvt.GetParameterType() == ParameterNameValueType::UInt16Type)
max = nvt.GetValueUInt16();
}
if (ii->FindNameValParam(MIN_PIXEL_INTENSITY_PARAM_NAME, nvt))
{
if (nvt.GetParameterType() == ParameterNameValueType::UInt16Type)
min = nvt.GetValueUInt16();
}
wchar_t buf[30];
FormatString2(buf, 30, L"[%d..%d]", min, max);
datHeader = buf;
datHeader += partialDatHeader;
}
}
break;
}
}
return StringUtils::ConvertWCSToMBS(datHeader);
}
void CHPMultiDataFileReaderTest::testRead()
{
affymetrix_calvin_data::ProbeSetMultiDataExpressionData ex;
affymetrix_calvin_data::ProbeSetMultiDataGenotypeData gn;
ParameterNameValueType param;
CHPMultiDataData data2;
CHPMultiDataFileReader reader;
reader.SetFilename("../data/CHP_MultiData_file");
reader.Read(data2);
CPPUNIT_ASSERT(data2.GetAlgName() == L"sig");
CPPUNIT_ASSERT(data2.GetAlgVersion() == L"1.0");
CPPUNIT_ASSERT(data2.GetArrayType() == L"test3");
CPPUNIT_ASSERT(data2.GetEntryCount(ExpressionMultiDataType) == 1);
CPPUNIT_ASSERT(data2.GetEntryCount(GenotypeMultiDataType) == 2);
// expression
CPPUNIT_ASSERT(data2.GetNumMetricColumns(ExpressionMultiDataType) == 0);
data2.GetExpressionEntry(ExpressionMultiDataType, 0, ex);
CPPUNIT_ASSERT_DOUBLES_EQUAL(ex.quantification, 10.0f, 0.0001f);
CPPUNIT_ASSERT(ex.name == "ex1");
// genotype
CPPUNIT_ASSERT(data2.GetNumMetricColumns(GenotypeMultiDataType) == 2);
CPPUNIT_ASSERT(data2.GetMetricColumnName(GenotypeMultiDataType, 0) == L"int");
CPPUNIT_ASSERT(data2.GetMetricColumnName(GenotypeMultiDataType, 1) == L"float");
data2.GetGenotypeEntry(GenotypeMultiDataType, 0, gn);
CPPUNIT_ASSERT(gn.call == 1);
CPPUNIT_ASSERT_DOUBLES_EQUAL(gn.confidence, 11.0f, 0.0001f);
CPPUNIT_ASSERT(gn.name == "gn1");
CPPUNIT_ASSERT(gn.metrics.size() == 2);
param = gn.metrics[0];
CPPUNIT_ASSERT(param.GetParameterType() == ParameterNameValueType::Int32Type);
CPPUNIT_ASSERT(param.GetValueInt32() == 1);
param = gn.metrics[1];
CPPUNIT_ASSERT(param.GetParameterType() == ParameterNameValueType::FloatType);
CPPUNIT_ASSERT_DOUBLES_EQUAL(param.GetValueFloat(), 2.0f, 0.00001f);
data2.GetGenotypeEntry(GenotypeMultiDataType, 1, gn);
CPPUNIT_ASSERT(gn.call == 2);
CPPUNIT_ASSERT_DOUBLES_EQUAL(gn.confidence, 22.0f, 0.0001f);
CPPUNIT_ASSERT(gn.name == "gn2");
CPPUNIT_ASSERT(gn.metrics.size() == 2);
param = gn.metrics[0];
CPPUNIT_ASSERT(param.GetParameterType() == ParameterNameValueType::Int32Type);
CPPUNIT_ASSERT(param.GetValueInt32() == 2);
param = gn.metrics[1];
CPPUNIT_ASSERT(param.GetParameterType() == ParameterNameValueType::FloatType);
CPPUNIT_ASSERT_DOUBLES_EQUAL(param.GetValueFloat(), 3.0f, 0.00001f);
}
void TestFileGenerator::WriteSmallDatFileWithReservedStringParameters()
{
int32_t rows = 4;
int32_t cols = 5;
int32_t pixelCount = rows*cols;
DATData data("small_dat_file_with_reserved_string_parameters");
data.SetPixelCount(pixelCount);
data.SetStatsCount(1);
data.SetArrayType(L"Hg_small");
data.SetPixelSize(0.71f);
data.SetScannerType(L"M10");
data.SetScannerID(L"main");
DateTime dt = DateTime::Parse(L"2005-12-25T11:12:13Z");
data.SetScanDate(dt);
data.SetRows(rows);
data.SetCols(cols);
std::string arrayId = "smellsliketeenspirit";
data.SetArrayId(arrayId);
ParameterNameValueType nvt;
nvt.SetName(L"fixedlen");
nvt.SetValueText(L"twenty-five", 25);
data.GetFileHeader()->GetGenericDataHdr()->AddNameValParam(nvt);
DATFileWriter writer(data);
Uint16Vector pixels;
pixels.reserve(pixelCount);
u_int16_t inten = 10;
for (int32_t i=0; i<pixelCount; ++i, ++inten)
{
if (inten > 46000)
{
inten = 0;
}
pixels.push_back(inten);
}
Uint16Vector stats;
stats.push_back(pixels[0]); // min
stats.push_back(pixels[19]); // max
writer.WriteStats(stats);
writer.WritePixels(pixels);
}
void TestFileGenerator::WriteCelFileWithADataSetWithZeroRows()
{
CelFileData data("small_cel_file_with_dataset_of_zero_rows");
data.SetIntensityCount(25);
data.SetStdDevCount(25);
data.SetPixelCount(25);
data.SetOutlierCount(0);
data.SetMaskCount(0);
data.SetRows(5);
data.SetCols(5);
data.SetArrayType(L"Hg-small");
data.SetAlgorithmName(L"Feature Extraction");
ParameterNameValueType nvt;
nvt.SetName(L"percentile");
nvt.SetValueFloat(0.75f);
data.AddAlgorithmParameter(nvt);
nvt.SetName(L"outlierlow");
nvt.SetValueFloat(1.004f);
data.AddAlgorithmParameter(nvt);
CelFileWriter* writer = new CelFileWriter(data);
FloatVector vInten;
FloatVector vStdev;
Int16Vector vPixels;
for (int i=0; i<25; ++i)
{
vInten.push_back(100.0f*i);
vStdev.push_back(.123f);
vPixels.push_back(25);
}
// Do some writing
writer->WriteIntensities(vInten);
writer->WriteStdDevs(vStdev);
writer->WritePixels(vPixels);
XYCoordVector outlier;
writer->WriteOutlierCoords(outlier);
XYCoordVector masked;
writer->WriteMaskCoords(masked);
delete writer;
}
void TestFileGenerator::WriteSmallCelFileWithAPartialDatHeaderTest()
{
CelFileData data("small_cel_file_partial_datheader");
// Write
ParameterNameValueType nvt;
GenericDataHeader datHdr;
datHdr.SetFileId(AffymetrixGuid::GenerateNewGuid());
datHdr.SetFileTypeId("affymetrix-calvin-scan-acquisition");
datHdr.SetFileCreationTime(L"2004-07-01T13:14:15Z");
nvt.SetName(L"affymetrix-partial-dat-header");
std::wstring datHeaderString = L" small_cel_file_partial_datheader:CLS=25 RWS=25 XIN=1 YIN=1 VE=0 0 05/19/05 02:45:59 ScannerID: ScannerTyp \x14 \x14 Hg-Small.1sq \x14 \x14 \x14 \x14 \x14 570 \x14 45.200001 \x14 0.340000 \x14 1.0900 \x14 3";
nvt.SetValueText(datHeaderString);
datHdr.AddNameValParam(nvt);
nvt.SetName(L"affymetrix-max-pixel-intensity");
nvt.SetValueUInt16(46001);
datHdr.AddNameValParam(nvt);
nvt.SetName(L"affymetrix-min-pixel-intensity");
nvt.SetValueUInt16(1);
datHdr.AddNameValParam(nvt);
// Add DAT GenericDataHeader as parent.
data.GetFileHeader()->GetGenericDataHdr()->AddParent(datHdr);
WriteRemaingSmallCelFileWithGridParameters(data);
}
void CELConversionUtilities::SetCellMargin(CelFileData &inFile, CCELFileData &outFile, int defaultValue)
{
ParameterNameValueType nvt;
if (inFile.FindAlgorithmParameter(L"CellMargin", nvt))
{
if (nvt.GetParameterType() == ParameterNameValueType::Int32Type)
{
outFile.GetHeader().SetMargin(nvt.GetValueInt32());
}
else if(nvt.GetParameterType() == ParameterNameValueType::AsciiType)
{
outFile.GetHeader().SetMargin(atoi(nvt.GetValueAscii().c_str()));
}
}
else
{
outFile.GetHeader().SetMargin(defaultValue);
}
}
void TestFileGenerator::WriteSmallDatFileWithGridAndSubgridsAndParameters()
{
std::string filename = "small_DAT_file_with_subgrids_and_parameters";
WriteDatFile(filename, L"Hg-small", 4, 5, true);
DATData data(filename);
AddGridAndSubgrids(data, 2.5, 10);
ParameterNameValueType nvt;
nvt.SetName(L"Zip Code");
nvt.SetValueInt32(95051);
data.AddGridAlignmentAlgorithmParameter(nvt);
nvt.SetName(L"County");
nvt.SetValueText(L"Santa Clara");
data.AddGridAlignmentAlgorithmParameter(nvt);
DATFileUpdater update(data);
update.Update();
}
