// build the read group library map
void Dedup_LowMem::buildReadGroupLibraryMap(SamFileHeader& header) {
rgidLibMap.clear();
numLibraries = 0;
std::map<std::string,uint32_t> libNameMap;
SamHeaderRecord * headerRecord = header.getNextRGRecord();
while(headerRecord != NULL) {
std::string ID = headerRecord->getTagValue("ID");
std::string LB = headerRecord->getTagValue("LB");
if ( ID.empty() ) {
std::string headerRecordString;
headerRecord->appendString(headerRecordString);
Logger::gLogger->error("Cannot find readGroup ID information in the header line %s",
headerRecordString.c_str());
}
if ( rgidLibMap.find(ID) != rgidLibMap.end() ) {
Logger::gLogger->error("The readGroup ID %s is not a unique identifier",ID.c_str());
}
if ( LB.empty() ) {
std::string headerRecordString;
headerRecord->appendString(headerRecordString);
Logger::gLogger->warning("Cannot find library information in the header line %s. Using empty string for library name",
headerRecordString.c_str());
}
if ( libNameMap.find( LB ) != libNameMap.end() ) {
rgidLibMap[ID] = libNameMap[LB];
}
else {
numLibraries = libNameMap.size()+1;
libNameMap[LB] = numLibraries;
rgidLibMap[ID] = numLibraries;
}
headerRecord = header.getNextRGRecord();
}
if (numLibraries > 0xff) {
Logger::gLogger->error("More than 255 library names are identified. Dedup_LowMem currently only allows up to 255 library names");
}
}
void parseOutRG(SamFileHeader& header, std::string& noRgPgString, SamFileHeader* newHeader)
{
noRgPgString.clear();
// strings for comparing if two RGs with same ID are the same.
static std::string prevString = "";
static std::string newString = "";
SamHeaderRecord* rec = header.getNextHeaderRecord();
while(rec != NULL)
{
if(rec->getType() == SamHeaderRecord::RG)
{
if(newHeader != NULL)
{
// This is an RG line.
// First check if this RG is already included in the new header.
SamHeaderRG* prevRG = newHeader->getRG(rec->getTagValue("ID"));
if(prevRG != NULL)
{
// This RG already exists, check that they are the same.
// If they are the same, there is nothing to do.
bool status = true;
prevString.clear();
newString.clear();
status &= prevRG->appendString(prevString);
status &= rec->appendString(newString);
if(prevString != newString)
{
// They are not identical, so report an error.
Logger::gLogger->error("Failed to add readgroup to header, "
"duplicate, but non-identical RG ID, %s",
rec->getTagValue("ID"));
}
}
else
{
// This RG does not exist yet, so add it to the new header.
if(!newHeader->addRecordCopy((SamHeaderRG&)(*rec)))
{
// Failed to add the RG, exit.
Logger::gLogger->error("Failed to add readgroup to header, %s",
newHeader->getErrorMessage());
}
}
}
}
else if(rec->getType() == SamHeaderRecord::PG)
{
if(newHeader != NULL)
{
// This is a PG line.
// First check if this PG is already included in the new header.
SamHeaderPG* prevPG = newHeader->getPG(rec->getTagValue("ID"));
if(prevPG != NULL)
{
// This PG already exists, check if they are the same.
// If they are the same, there is nothing to do.
bool status = true;
prevString.clear();
newString.clear();
status &= prevPG->appendString(prevString);
status &= rec->appendString(newString);
if(prevString != newString)
{
// They are not identical, ignore for now.
// TODO: change the ID, and add it.
Logger::gLogger->warning("Warning: dropping duplicate, "
"but non-identical PG ID, %s",
rec->getTagValue("ID"));
}
}
else
{
// This PG does not exist yet, so add it to the new header.
if(!newHeader->addRecordCopy((SamHeaderPG&)(*rec)))
{
// Failed to add the PG, exit.
Logger::gLogger->error("Failed to add PG to header, %s",
newHeader->getErrorMessage());
}
}
}
}
else
{
rec->appendString(noRgPgString);
}
rec = header.getNextHeaderRecord();
}
// Append the comments.
header.appendCommentLines(noRgPgString);
}
void parseOutRG(SamFileHeader& header, std::string& noRgPgString, SamFileHeader* newHeader, bool ignorePI)
{
noRgPgString.clear();
// strings for comparing if two RGs with same ID are the same.
static std::string prevString = "";
static std::string newString = "";
SamHeaderRecord* rec = header.getNextHeaderRecord();
while(rec != NULL)
{
if(rec->getType() == SamHeaderRecord::RG)
{
if(newHeader != NULL)
{
// This is an RG line.
// First check if this RG is already included in the new header.
SamHeaderRG* prevRG = newHeader->getRG(rec->getTagValue("ID"));
if(prevRG != NULL)
{
// This RG already exists, check that they are the same.
// If they are the same, there is nothing to do.
bool status = true;
prevString.clear();
newString.clear();
status &= prevRG->appendString(prevString);
status &= rec->appendString(newString);
if(prevString != newString)
{
if(!ignorePI)
{
Logger::gLogger->error("Failed to add readgroup to "
"header, duplicate, but "
"non-identical RG ID, %s\n"
"prev:\t(%s)\nnew:\t(%s)",
rec->getTagValue("ID"),
prevString.c_str(),
newString.c_str());
}
else
{
// Check for a PI string.
size_t prevPIStart = prevString.find("PI:");
size_t newPIStart = newString.find("PI:");
// If they are both npos, then PI was not found
// so fail.
if((prevPIStart == std::string::npos) &&
(newPIStart == std::string::npos))
{
// They are not identical, so report an error.
Logger::gLogger->error("Failed to add readgroup"
" to header, duplicate,"
" but non-identical RG"
" ID, %s\n"
"prev:\t(%s)\nnew:\t(%s)",
rec->getTagValue("ID"),
prevString.c_str(),
newString.c_str());
}
else
{
// PI found in one or both strings.
size_t prevPIEnd;
size_t newPIEnd;
if(prevPIStart == std::string::npos)
{
// new string has PI, so compare to the start of that.
prevPIStart = newPIStart;
prevPIEnd = newPIStart;
}
else
{
prevPIEnd = prevString.find('\t', prevPIStart) + 1;
}
if(newPIStart == std::string::npos)
{
// new string has PI, so compare to the start of that.
newPIStart = prevPIStart;
newPIEnd = newPIStart;
}
else
{
newPIEnd = newString.find('\t', newPIStart) + 1;
}
// Compare before PI.
if((newString.compare(0, newPIStart, prevString, 0, prevPIStart) != 0) ||
(newString.compare(newPIEnd, std::string::npos, prevString,
prevPIEnd, std::string::npos) != 0))
{
// They are not identical, so report an error.
Logger::gLogger->error("Failed to add readgroup to header, "
"duplicate, but non-identical RG ID, %s, "
"even when ignoring PI\n"
"prev:\t(%s)\nnew:\t(%s)",
rec->getTagValue("ID"),
prevString.c_str(),
newString.c_str());
}
else
{
Logger::gLogger->warning("Warning: ignoring non-identical PI field "
"for RG ID, %s",
rec->getTagValue("ID"));
}
}
}
}
}
else
{
// This RG does not exist yet, so add it to the new header.
if(!newHeader->addRecordCopy((SamHeaderRG&)(*rec)))
{
// Failed to add the RG, exit.
Logger::gLogger->error("Failed to add readgroup to header, %s",
newHeader->getErrorMessage());
}
}
}
}
else if(rec->getType() == SamHeaderRecord::PG)
{
if(newHeader != NULL)
{
// This is a PG line.
// First check if this PG is already included in the new header.
SamHeaderPG* prevPG = newHeader->getPG(rec->getTagValue("ID"));
if(prevPG != NULL)
{
// This PG already exists, check if they are the same.
// If they are the same, there is nothing to do.
bool status = true;
prevString.clear();
newString.clear();
status &= prevPG->appendString(prevString);
status &= rec->appendString(newString);
if(prevString != newString)
{
// They are not identical, ignore for now.
// TODO: change the ID, and add it.
Logger::gLogger->warning("Warning: dropping duplicate, "
"but non-identical PG ID, %s",
rec->getTagValue("ID"));
}
}
else
{
// This PG does not exist yet, so add it to the new header.
if(!newHeader->addRecordCopy((SamHeaderPG&)(*rec)))
{
// Failed to add the PG, exit.
Logger::gLogger->error("Failed to add PG to header, %s",
newHeader->getErrorMessage());
}
}
}
}
else
{
rec->appendString(noRgPgString);
}
rec = header.getNextHeaderRecord();
}
// Append the comments.
header.appendCommentLines(noRgPgString);
}
