void
affx::TsvFileDiff::open_residuals(const std::string& filename,
affx::TsvFile& tsv1,
affx::TsvFile& tsv2)
{
// no file
if (filename=="") {
return;
}
//
m_residuals_tsv=new affx::TsvFile();
//
for (int clvl=0;clvl<tsv1.getLevelCount();clvl++) {
for (int cidx=0;cidx<tsv1.getColumnCount(clvl);cidx++) {
std::string cname=tsv1.getColumnName(clvl,cidx); // +"_res";
m_residuals_tsv->defineColumn(clvl,cidx,cname);
}
}
//
m_residuals_tsv->addHeader("apt-file-type","tsv-diff-residuals");
m_residuals_tsv->addHeader("tsv-diff-input-file1",tsv1.getFileName());
m_residuals_tsv->addHeader("tsv-diff-input-file2",tsv2.getFileName());
m_residuals_tsv->addHeader("file-guid",affxutil::Guid::GenerateNewGuid());
//
if (m_residuals_tsv->getLevelCount()==1) {
m_residuals_tsv->writeTsv_v1(filename);
} else {
m_residuals_tsv->writeTsv_v2(filename);
}
}
static void WriteHeader(MetabolizerPhenotypingEngine *eng, affx::TsvFile &tsv, int idx, std::ofstream &str)
{
// Output the header
str << "# For research use only. Not for diagnostic purposes." << std::endl;
str << "#%report-guid=" << affxutil::Guid::GenerateNewGuid() << std::endl;
str << "#%Program=" << MET_PROGRAM_NAME << std::endl;
str << "#%Version=" << MET_PROGRAM_VERSION << std::endl;
str << "#%Date=" << Util::getTimeStamp() << std::endl;
str << "#%MetabolizerFile=" << eng->getOpt("metabolizer-file") << std::endl;
str << "#%AlleleFile=" << eng->getOpt("allele-file") << std::endl;
for (int i=0; i<(int)eng->PhenotypeCallDescs().size(); i++)
str << "#%PhenotypeCallDesc=" << eng->PhenotypeCallDescs()[i] << std::endl;
for (int i=0; i<(int)eng->UserInformation().size(); i++)
str << "#%Info=" << eng->UserInformation()[i] << std::endl;
str << "Index" << "\t"
<< "CHP File" << "\t"
<< "Gene" << "\t"
<< "Phenotype Call" << "\t"
<< "Gene Activity" << "\t"
<< "Known Call" << "\t"
<< "Unknown Call" << "\t"
<< "Interpretation Code";
int n = tsv.getColumnCount(0);
for (int i=idx; i<n; i++)
{
std::string name = tsv.getColumnName(0, i);
str << "\t" << name;
}
str << std::endl;
}
void
affx::TsvFileDiff::p_diff_1line(affx::TsvFile& tsv1, affx::TsvFile& tsv2)
{
std::string val1,val2;
int tsv1_clvl=tsv1.lineLevel();
int tsv1_cidx_max=tsv1.getColumnCount(tsv1_clvl);
//
p_linenums(tsv1,tsv2);
printf("!");
//
for (int cidx=0;cidx<tsv1_cidx_max;cidx++) {
if (cidx!=0) {
printf("\t");
}
printf("\t");
tsv1.get(tsv1_clvl,cidx,val1);
tsv2.get(tsv1_clvl,cidx,val2);
if (val1==val2) {
printf("%s",val1.c_str());
}
else {
printf("'%s'/'%s'",val1.c_str(),val2.c_str());
}
}
printf("\n");
}
/**
* Check whether a clf file is sequential.
*
* @param clfTsv Clf file tsv object.
* @return bool Returns true if the file is sequential, else false.
*/
bool SequentialClfFile::isSequential(affx::TsvFile& clfTsv)
{
std::string rowsValue, colsValue, sequentialValue;
clfTsv.headersFindNext("rows", rowsValue);
clfTsv.headersFindNext("cols", colsValue);
clfTsv.headersFindNext("sequential", sequentialValue);
// Require rows, cols, and a sequential value of 1.
if (rowsValue.empty() || colsValue.empty() || sequentialValue != "1")
return false;
// Save rows, cols.
m_Rows = Convert::toInt(rowsValue);
m_Cols = Convert::toInt(colsValue);
return true;
}
static int FindStartingAttributeColumnIndex(affx::TsvFile &tsv)
{
int n = tsv.getColumnCount(0);
int idx = n;
for (int i=0; i<n; i++)
{
std::string name = tsv.getColumnName(0, i);
if (name == "Override Comment")
{
idx = i+1;
break;
}
}
return idx;
}
int
affx::TsvFileDiff::diffHeaders(affx::TsvFile& tsv1, affx::TsvFile& tsv2)
{
int start_diff_cnt=m_diff_cnt;
std::string key1,val1,key2,val2;
// do the diff in three passes.
// 1 = changed values
tsv1.headersBegin();
while (tsv1.headersNext(key1,val1)==affx::TSV_OK) {
if (tsv2.getHeader(key1,val2)==affx::TSV_OK) {
if (val1==val2) {
if (m_opt_print_same) {
if (p_inc()) {
printf(" #%%%s=%s\n",key1.c_str(),val1.c_str());
}
}
}
else { // !=
m_diff_cnt++;
if (p_inc()) {
printf("-#%%%s=%s\n",key1.c_str(),val1.c_str());
printf("+#%%%s=%s\n",key1.c_str(),val2.c_str());
}
}
}
}
// 2 = headers in tsv1 not in tsv2
tsv1.headersBegin();
while (tsv1.headersNext(key1,val1)==affx::TSV_OK) {
if (tsv2.getHeader(key1,val2)!=affx::TSV_OK) {
m_diff_cnt++;
if (p_inc()) {
printf("-#%%%s=%s\n",key1.c_str(),val1.c_str());
}
}
}
// 3 = headers in tsv2 not in tsv1
tsv2.headersBegin();
while (tsv2.headersNext(key2,val2)==affx::TSV_OK) {
if (tsv1.getHeader(key2,val1)!=affx::TSV_OK) {
m_diff_cnt++;
if (p_inc()) {
printf("+#%%%s=%s\n",key2.c_str(),val2.c_str());
}
}
}
//
return m_diff_cnt-start_diff_cnt;
}
void affx::TsvFileDiff::p_line(affx::TsvFile& tsv) {
int clvl=tsv.lineLevel();
int cidx_max=tsv.getColumnCount(clvl);
std::string val;
//
for (int i=0;i<clvl;i++) {
printf("\t");
}
//
for (int cidx=0;cidx<cidx_max;cidx++) {
if (cidx!=0) {
printf("\t");
}
printf("\t");
tsv.get(clvl,cidx,val);
printf("%s",val.c_str());
}
printf("\n");
}
static void OpenAlleleFile(MetabolizerPhenotypingEngine *eng, affx::TsvFile &tsv, MetabolizerResultType &result, int &sampleIdx, int &ncols, std::vector<std::string> &sampleCols)
{
std::string translationTableFile = eng->getOpt("allele-file");
tsv.m_optQuoteChar1 = 0;
tsv.m_optQuoteChar2 = 0;
if (tsv.open(translationTableFile) != affx::TSV_OK)
{
std::string err = "Unable to open the metabolizer bin file.";
Err::errAbort(err);
}
tsv.bind(0, "Index", &result.index, affx::TSV_BIND_REQUIRED);
tsv.bind(0, "CHP Filename", &result.chpFile, affx::TSV_BIND_REQUIRED);
tsv.bind(0, "Gene", &result.gene, affx::TSV_BIND_REQUIRED);
tsv.bind(0, "Known Call", &result.knownCall, affx::TSV_BIND_REQUIRED);
tsv.bind(0, "Unknown Call", &result.unknownCall, affx::TSV_BIND_REQUIRED);
tsv.bind(0, "Interpretation Code", &result.code, affx::TSV_BIND_REQUIRED);
sampleIdx = FindStartingAttributeColumnIndex(tsv);
ncols = tsv.getColumnCount(0);
sampleCols.resize(ncols-sampleIdx);
for (int i=sampleIdx; i<ncols; i++)
tsv.bind(0, i, &sampleCols[i-sampleIdx], affx::TSV_BIND_REQUIRED);
}
int
affx::TsvFileDiff::diffData(affx::TsvFile& tsv1, affx::TsvFile& tsv2)
{
int start_diff_cnt=m_diff_cnt;
std::string val1,val2;
//
open_residuals(m_residuals_filename,tsv1,tsv2);
//
tsv1.rewind();
tsv2.rewind();
//
tsv1.nextLine();
tsv2.nextLine();
//
while ((!tsv1.eof())||(!tsv2.eof())) {
//
int tsv1_clvl=tsv1.lineLevel();
int tsv2_clvl=tsv2.lineLevel();
//
while (((tsv1_clvl>tsv2_clvl)||tsv2.eof())&&(!tsv1.eof())) {
if (p_inc()) {
p_linenums(tsv1,tsv2);
printf("-");
p_line(tsv1);
}
m_diff_cnt+=tsv1.getColumnCount(tsv1_clvl);
tsv1.nextLine();
tsv1_clvl=tsv1.lineLevel();
}
//
while (((tsv2_clvl>tsv1_clvl)||tsv1.eof())&&(!tsv2.eof())) {
if (p_inc()) {
p_linenums(tsv1,tsv2);
printf("+");
p_line(tsv2);
}
m_diff_cnt+=tsv2.getColumnCount(tsv2_clvl);
tsv2.nextLine();
tsv2_clvl=tsv2.lineLevel();
}
//
if ((tsv1_clvl>=0)&&(tsv2_clvl>=0)) {
int tsv1_cidx_max=tsv1.getColumnCount(tsv1_clvl);
int tsv2_cidx_max=tsv2.getColumnCount(tsv2_clvl);
//
int cidx_max;
if (tsv1_cidx_max<tsv2_cidx_max) {
cidx_max=tsv2_cidx_max;
} else {
cidx_max=tsv1_cidx_max;
}
// count diffs on the line
int line_diff_cnt=0;
double d1,d2;
for (int cidx=0;cidx<cidx_max;cidx++) {
tsv1.get(tsv1_clvl,cidx,val1);
tsv2.get(tsv2_clvl,cidx,val2);
// the same?
if (val1==val2) {
// record the base value
if (m_residuals_tsv!=NULL) {
// numeric?
if (tsv1.get(tsv1_clvl,cidx,d1)==affx::TSV_OK) {
// put a zero
m_residuals_tsv->set(tsv1_clvl,cidx,"0");
} else {
// put the value
m_residuals_tsv->set(tsv1_clvl,cidx,val1);
}
}
}
// Not string equal -- check for numeric diff
else if ((tsv1.get(tsv1_clvl,cidx,d1)==affx::TSV_OK) &&
(tsv2.get(tsv2_clvl,cidx,d2)==affx::TSV_OK)) {
double d_diff=d1-d2;
// output the result...
if (m_residuals_tsv!=NULL) {
m_residuals_tsv->set(tsv1_clvl,cidx,d_diff);
}
// if small, it isnt a diff
if (!((m_opt_max_diff>0.0) && (fabs(d_diff)<=m_opt_max_diff))) {
// too big -- mark it down as a diff.
m_diff_cnt++;
line_diff_cnt++;
//
}
}
// normal text diff
else {
if (m_residuals_tsv!=NULL) {
m_residuals_tsv->set(tsv1_clvl,cidx,"'"+val1+"'/'"+val2+"'");
}
}
}
// output the result line
if (m_residuals_tsv!=NULL) {
m_residuals_tsv->writeLevel(tsv1_clvl);
}
// print unchanged lines?
if ((m_opt_print_same || (line_diff_cnt!=0)) &&
(p_inc())) {
//
if (line_diff_cnt==0) {
p_linenums(tsv1,tsv2);
printf(" ");
p_line(tsv1);
}
else if (m_opt_print_format==TsvFileDiff::FMT_1) {
p_diff_1line(tsv1,tsv2);
}
else if (m_opt_print_format==TsvFileDiff::FMT_2) {
p_diff_2line(tsv1,tsv2);
}
else {
Err::errAbort("Bad format");
}
}
}
//
tsv1.nextLine();
tsv2.nextLine();
}
close_residuals();
return m_diff_cnt-start_diff_cnt;
}
void
affx::TsvFileDiff::p_linenums(affx::TsvFile& tsv1,affx::TsvFile& tsv2)
{
p_linenums(tsv1.lineNumber(),tsv2.lineNumber());
}