ReadData.cc Source File

00001 #include "Exceptions.hh"
00002 #include "ReadData.hh"
00003 
00004 #include <stdlib.h> // for atol
00005 
00010 namespace libSlice
00011 {
00012   using std::string;
00013 
00015 
00024 ReadData::ReadData(const char * id, 
00025                    const char * seqname,
00026                    const char * dir, 
00027                    const char * chem)
00028 {
00029   if (!id)
00030   {
00031     throw Exception("Cannot have NULL read id", "ReadData::ReadData");
00032   }
00033 
00034   m_id = id;
00035 
00036   m_dir     = (dir)     ? dir     : "0";
00037   m_chem    = (chem)    ? chem    : "";
00038   m_seqname = (seqname) ? seqname : "";
00039 
00040   m_clrleft  = m_seqleft = 0;
00041   m_clrright = m_seqright = 0;
00042 
00043   parseDirection();
00044 }
00045 
00047 
00054 void ReadData::addClrData(const char * left,
00055                           const char * right)
00056 {
00057   setSeqData((left)  ? atol(left)  : 0,
00058              (right) ? atol(right) : 0);
00059 }
00060 
00061 void ReadData::setSeqData(long int seqleft,
00062                           long int seqright)
00063 {
00064   m_seqleft  = seqleft;
00065   m_seqright = seqright;
00066 
00067   m_clrleft  = (m_rc) ? m_seqright : m_seqleft;
00068   m_clrright = (m_rc) ? m_seqleft  : m_seqright;
00069 
00070   if (m_clrright < m_clrleft)
00071   {
00072     throw Exception("Invalid Clrdata for id " + m_id,
00073                         "ReadData::addClrData");
00074   }
00075 }
00076 
00077 void ReadData::setClrData(long int clrleft,
00078                           long int clrright)
00079 {
00080   if (m_rc)
00081   {
00082     setSeqData(clrright, clrleft);
00083   }
00084   else
00085   {
00086     setSeqData(clrleft, clrright);
00087 
00088   }
00089 }
00090 
00091 
00093 
00098 void ReadData::parseDirection()
00099 {
00100   if (m_dir == "0" || m_dir[0] == 'F')
00101   {
00102     m_rc = 0;
00103   }
00104   else if (m_dir == "1" || m_dir[0] == 'R')
00105   {
00106     m_rc = 1;
00107   }
00108   else
00109   {
00110     throw Exception("Read Direction is unknown in " + m_dir,
00111                         "ReadData::ReadData");
00112   }
00113 }
00114 
00116 
00119 char ReadData::isReverseCompliment() const
00120 {
00121   return m_rc;
00122 }
00123 
00125 
00131 void ReadData::printXML(ostream & out, const string & prefix) const
00132 {
00133   out << prefix 
00134       << "<Read Id=\""    << m_id
00135       << "\" Seqname=\""  << m_seqname
00136       << "\" Dir=\""      << m_dir
00137       << "\" Chem=\""     << m_chem 
00138       << "\">" << endl;
00139 
00140   if (m_seqright)
00141   {
00142     out << prefix 
00143         << "  <Clr Left=\"" << m_seqleft
00144         << "\" Right=\""    << m_seqright
00145         << "\" />" << endl;
00146   }
00147 
00148   if (m_leftnuc.length())
00149   {
00150     out << prefix << "  <TrimmedSeq position=\"left\">"            << endl
00151         << prefix << "    <Nuc>"     << m_leftnuc  << "</Nuc>"     << endl
00152         << prefix << "    <Qualval>" << m_leftqual << "</Qualval>" << endl
00153         << prefix << "  </TrimmedSeq>"                             << endl;
00154   }
00155 
00156   if (m_rightnuc.length())
00157   {
00158     out << prefix << "  <TrimmedSeq position=\"right\">"            << endl
00159         << prefix << "    <Nuc>"     << m_rightnuc  << "</Nuc>"     << endl
00160         << prefix << "    <Qualval>" << m_rightqual << "</Qualval>" << endl
00161         << prefix << "  </TrimmedSeq>"                              << endl;
00162   }
00163 
00164   out << prefix << "</Read>" << endl;
00165 }
00166 
00168 
00170 void ReadData::reverseCompliment()
00171 {
00172   if (m_dir == "Forward")
00173   {
00174     m_dir = "Reversed";
00175   }
00176   else if (m_dir == "0")
00177   {
00178     m_dir = "1";
00179   }
00180   else if (m_dir == "Reversed")
00181   {
00182     m_dir = "Forward";
00183   }
00184   else if (m_dir == "1")
00185   {
00186     m_dir = "0";
00187   }
00188  
00189   long swap = m_seqleft;
00190   m_seqleft = m_seqright;
00191   m_seqright = swap;
00192 }
00193 
00195 
00203 void ReadData::resetId(const string & id)
00204 {
00205   m_id = id;
00206 }
00207 
00208 
00209 
00211 const string & ReadData::getId() const
00212 {
00213   return m_id;
00214 }
00215 
00217 const string & ReadData::getSeqname() const
00218 {
00219   return m_seqname;
00220 }
00221 
00223 const string & ReadData::getChem() const
00224 {
00225   return m_chem;
00226 }
00227 
00229 long ReadData::getSeqleft() const
00230 {
00231   return m_seqleft;
00232 }
00233 
00235 long ReadData::getSeqright() const
00236 {
00237   return m_seqright;
00238 }
00239 
00241 long ReadData::getClrleft() const
00242 {
00243   return m_clrleft;
00244 }
00245 
00247 long ReadData::getClrright() const
00248 {
00249   return m_clrright;
00250 }
00251 
00253 void ReadData::addNucData(const char * buffer, bool isLeft)
00254 {
00255   if (buffer)
00256   {
00257     if (isLeft)
00258     {
00259       m_leftnuc += buffer;
00260     }
00261     else
00262     {
00263       m_rightnuc += buffer;
00264     }
00265   }
00266 }
00267 
00269 void ReadData::addQualData(const char * buffer, bool isLeft)
00270 {
00271   if (buffer)
00272   {
00273     if (isLeft)
00274     {
00275       m_leftqual += buffer;
00276     }
00277     else
00278     {
00279       m_rightqual += buffer;
00280     }
00281   }
00282 }
00283 
00285 const string & ReadData::getRawNucData(bool getLeft) const
00286 {
00287   if (getLeft) return m_leftnuc;
00288   return m_rightnuc;
00289 }
00290 
00291 
00293 const string & ReadData::getRawQualData(bool getLeft) const
00294 {
00295   if (getLeft) return m_leftqual;
00296   return m_rightqual;
00297 }
00298 
00300 void ReadData::validateSeqdata() const
00301 {
00302   //TODO: Implement this
00303 
00304 }
00305 
00306 
00308 
00311 const string ReadData::getNucData(bool getLeft) const
00312 {
00313   validateSeqdata();
00314 
00315   string retval;
00316 
00317   if (getLeft)
00318   {
00319     retval = m_leftnuc;
00320 
00321     if (!retval.size())
00322     {
00323       for (long i = 0; i < m_clrleft-1; i++)
00324       {
00325         retval += "?";
00326       }
00327     }
00328   }
00329   else
00330   {
00331     retval = m_rightnuc;
00332     
00333     // Can't pad here, because length is unknown
00334   }
00335 
00336   return retval;
00337 }
00338 
00340 
00343 const string ReadData::getQualData(bool getLeft) const
00344 {
00345   validateSeqdata();
00346 
00347   string retval;
00348 
00349   if (getLeft)
00350   {
00351     retval = m_leftqual;
00352 
00353     if (!retval.size())
00354     {
00355       bool isFirst = true;
00356       for (long i = 0; i < m_clrleft-1; i++)
00357       {
00358         if (!isFirst) retval += " ";
00359         retval += "00";
00360         isFirst = false;
00361       }
00362     }
00363   }
00364   else
00365   {
00366     retval = m_rightqual;
00367     
00368     // Can't pad here, because length is unknown
00369   }
00370 
00371   return retval;
00372 }
00373 
00374 
00376 void ReadData::flushTrimmedSeq()
00377 {
00378   m_leftnuc   = "";
00379   m_leftqual  = "";
00380 
00381   m_rightnuc  = "";
00382   m_rightqual = "";
00383 }
00384 
00386 void ReadData::flushTrimmedSeqSide(bool flushLeft)
00387 {
00388   if (flushLeft)
00389   {
00390     m_leftnuc = "";
00391     m_leftqual = "";
00392   }
00393   else
00394   {
00395     m_rightnuc = "";
00396     m_rightqual = "";
00397   }
00398 }
00399 
00400 
00402 
00404 ReadManager::ReadManager()
00405 {
00406   m_readIterator = m_readMap.end();
00407 }
00408 
00410 ReadManager::~ReadManager()
00411 {
00412   ReadMap::iterator i;
00413 
00414   for (i =  m_readMap.begin();
00415        i != m_readMap.end();
00416        i++)
00417   {
00418     delete i->second;
00419   } 
00420 }
00421 
00423 
00429 ReadData * ReadManager::addReadData(const char * id,
00430                                     const char * seqname,
00431                                     const char * dir,
00432                                     const char * chem)
00433 {
00434   ReadData * newRead = new ReadData(id, seqname, dir, chem);
00435 
00436   insertRead(newRead);
00437 
00438   return newRead;
00439 }
00440 
00442 ReadData * ReadManager::getNextRead()
00443 {
00444   ReadData * retval = NULL;
00445   
00446   if (m_readIterator != m_readMap.end())
00447   {
00448     retval = m_readIterator->second;
00449     m_readIterator++;
00450   }
00451 
00452   return retval;
00453 }
00454 
00456 void ReadManager::resetReadIterator()
00457 {
00458   m_readIterator = m_readMap.begin();
00459 }
00460 
00462 
00465 ReadData * ReadManager::find(const string & readId) const
00466 {
00467   ReadData * retval = NULL;
00468   ReadMap::const_iterator i = m_readMap.find(readId);
00469 
00470   if (i != m_readMap.end())
00471   {
00472     retval = i->second;
00473   }
00474 
00475   return retval;
00476 }
00477 
00479 int ReadManager::size() const
00480 {
00481   return m_readMap.size();
00482 }
00483 
00485 void ReadManager::removeRead(const string & id)
00486 {
00487   m_readMap.erase(id);
00488 }
00489 
00491 
00494 void ReadManager::insertRead(ReadData * rd)
00495 {
00496   const string & id = rd->getId();
00497 
00498   if (find(id))
00499   {
00500     throw Exception(id + " is already in readMap", 
00501                         "ReadManager::insertRead");
00502   }
00503 
00504   m_readMap[id] = rd;
00505 }
00506 
00508 
00512 void ReadManager::resetReadId(ReadData * rd, const string & id)
00513 {
00514   const string & origId = rd->getId();
00515 
00516   if (find(origId))
00517   {
00518     removeRead(origId);
00519     rd->resetId(id);
00520     insertRead(rd);
00521   }
00522 }
00523 
00525 void ReadManager::printXML(ostream & out, const string & prefix) const
00526 {
00527   out << prefix << "<ReadCollection>" << endl;
00528 
00529   ReadMap::const_iterator i;
00530   for (i =  m_readMap.begin();
00531        i != m_readMap.end();
00532        i++)
00533   {
00534     i->second->printXML(out, prefix + "  ");
00535   }
00536 
00537   out << prefix << "</ReadCollection>" << endl;
00538 }
00539 
00540 } // namespace