// Подготовка входных данных (формируем вектор хромосом и имен из мультимэпа особей)
ErrCode Crossover::get_DataIn_from_Individuals(DataIn& dataIn, Individuals& individuals)
{
Individuals::iterator itIndividual = individuals.begin();
size_t nIndividual = individuals.size();
int nChromoSector_in_Chromoset = sampleIndividual.nChromoSector_in_Chromoset;
dataIn.chromoSectors.resize(nIndividual*nChromoSector_in_Chromoset);
dataIn.expressions.resize(nIndividual);
for (int i = 0; i < nIndividual; ++i)
{
memcpy(&(dataIn.chromoSectors[i*nChromoSector_in_Chromoset]), &(itIndividual->second.chromoSectors[0]), sampleIndividual.nByte_in_Chromoset);
dataIn.expressions[i] = itIndividual->second.expression;
++itIndividual;
}
return 0;
}
ErrCode Crossover::get_crossedIndividuals(Individuals& crossedIndividuals, Individuals& individuals, const Device& device)
{
ErrCode err = 0;
// Параметры
int nThread_in_Warp = device.nMaxThread_in_Warp;
int nThread_in_Group = nThread_in_Warp * nWarp_in_Group;
int nThread = nThread_in_Group * nGroup_in_CU * device.max_compute_units;
// Локальные
int loc_sample_ByteSize = sampleIndividual.nByte_in_Chromoset;
int loc_sampleInvert_ByteSize = sampleIndividual.nByte_in_Chromoset;
int nIndividual_in_Loc = (device.LDS_ByteSize / nGroup_in_CU - loc_sample_ByteSize - loc_sampleInvert_ByteSize) / (sizeof(int)+sampleIndividual.nByte_in_Chromoset);
int loc_indexes_ByteSize = nIndividual_in_Loc * sizeof(int);
int loc_Individuals_ByteSize = nIndividual_in_Loc * sampleIndividual.nByte_in_Chromoset;
int loc_ByteSize = loc_sample_ByteSize + loc_sampleInvert_ByteSize + loc_indexes_ByteSize + loc_Individuals_ByteSize;
Editor editor;
err = editor.import_(kernel_filePath);
if (err != 0) return err;
editor.replace("$$EXPRESSION", expression);
calc_nParent();
editor.replace("$$N_PARENT", intToString(nParent));
editor.replace("$$N_CHROMOSECTOR_IN_CHROMOSET", intToString(sampleIndividual.chromoSectors.size()));
editor.replace("$$N_CHROMOSET_IN_LOC", intToString(nIndividual_in_Loc));
editor.replace("$$N_CHROMOSET_IN_GLOB", intToString(individuals.size()-10));
int min_rating = (*(--individuals.end())).second.rating;
editor.replace("$$START_RATING", intToString(min_rating));
editor.replace("$$N_GLOBAL_CICLES", intToString(nGlobal_Cicles));
editor.replace("$$N_LOCAL_CICLES", intToString(nLocal_Cicles));
editor.replace("$$K_RATING", floatToString(kRating));
editor.replace("$$KOEF_A", intToString(15678));
editor.replace("$$KOEF_C", intToString(34302));
editor.expand("$$EXTRACT_N_PARENT", nParent);
editor.export_(kernel_filePath + KERNEL_FILE_SUFFIX);
std::string kernel_Source;
editor.get_string(kernel_Source);
editor.clear();
Computer computer;
computer.set_Device(device);
computer.set_Kernel_Name("crossing");
computer.set_Kernel_Source(kernel_Source);
computer.set_nThread_in_Group(nThread_in_Group);
computer.set_nThread(nThread);
// Строим (компиляция ядра, создание контекста...)
err = computer.build_kernel();
if (err != 0) return err;
DataIn dataIn;
err = get_DataIn_from_Individuals(dataIn, individuals);
// Входа
computer.add_In(dataIn.chromoSectors);
computer.add_In(sampleIndividual.chromoSectors);
computer.add_In(sampleIndividual.invertChromoSectors);
computer.add_Local(loc_ByteSize);
DataOut dataOut(nThread, nParent);
// Выхода
computer.add_Out(dataOut.glob_indexes);
computer.add_Out(dataOut.ratings);
computer.add_Out(dataOut.c11s);
computer.add_Out(dataOut.c01s);
// Запускаем
computer.compute();
// Восстанавливаем особей
err = get_Individuals_from_DataOut(crossedIndividuals, dataOut);
computeTime = computer.get_computeTime();
// Освобождение ресурсов
computer.clear();
return 0;
}
// mine(): main function for GERMLINE
void GERMLINE::mine( string params, string prefix )
{
PolymorphicIndividualsExtractor * pie = inputManager.getPie();
inputManager.getIndividuals(prefix);
if ( ! pie->valid() ) return;
string out = prefix;
num_samples = 0;
num_matches = 0;
pie->loadInput();
MatchesBuilder mb( pie );
ofstream fout( ( out + ".log" ).c_str() );
fout << setw(65) << setfill('-') << ' ' << endl << setfill(' ');
fout << " Welcome to GERMLINE, a tool for detecting long segments shared" << endl;
fout << " by descent between pairs of individuals in near-linear time." << endl;
fout << endl;
fout << " For more details, please see the paper [ PMID: 18971310 ]" << endl;
fout << " or the web-site [ http://www.cs.columbia.edu/~gusev/germline/ ]" << endl;
fout << endl;
fout << " GERMLINE was coded by Alexander Gusev and collaborators in " << endl;
fout << " Itsik Pe'er's Computational Biology Lab at Columbia University" << endl;
fout << setw(65) << setfill('-') << ' ' << endl << setfill(' ');
if ( BINARY_OUT ) MATCH_FILE.open( ( out + ".bmatch" ).c_str() , ios::binary );
else MATCH_FILE.open( ( out + ".match" ).c_str() );
fout << params << endl;
fout << setw(65) << setfill('-') << ' ' << endl << setfill(' ');
//fout << setw(50) << left << "Minimum match length: " << MIN_MATCH_LEN << " cM" << endl;
fout << setw(50) << "Allowed mismatching bits: " << MAX_ERR_HOM << " " << MAX_ERR_HET << endl;
fout << setw(50) << "Word size: " << MARKER_SET_SIZE << endl;
if ( ROI )
fout << setw(50) << "Target region: " << ALL_SNPS.getROIStart().getSNPID() << " - " << ALL_SNPS.getROIEnd().getSNPID() << endl;
else
fout << setw(50) << "Target region: " << "all" << endl;
time_t timer[2]; time( &timer[0] );
if ( DEBUG ) cout << "DEBUG MODE ON" << endl;
if ( ROI )
{
ALL_SNPS.beginChromosome();
num_sets = (long)ceil((float)ALL_SNPS.currentSize()/(float)MARKER_SET_SIZE);
mb.buildMatches();
ALL_SAMPLES.freeMatches();
ALL_SAMPLES.freeMarkers();
}
else
{
int temp_i = 0;
for ( ALL_SNPS.beginChromosome() ; ALL_SNPS.moreChromosome() ; ALL_SNPS.nextChromosome() )
{
num_sets = (long)ceil((float)ALL_SNPS.currentSize()/(float)MARKER_SET_SIZE);
mb.buildMatches();
if ( !SILENT ) cout << "Matches completed ... freeing memory" << endl;
ALL_SAMPLES.freeMatches();
ALL_SAMPLES.freeMarkers();
}
}
//time( &timer[1] );
fout << setw(50) << "Total IBD segments: " << num_matches << endl;
//fout << setw(50) << "Total runtime (sec): " << difftime( timer[1] , timer[0] ) << endl;
fout.close();
MATCH_FILE.close();
if ( BINARY_OUT )
{
ofstream bmid_out( ( out + ".bmid" ).c_str() );
ALL_SNPS.print( bmid_out );
bmid_out.close();
ofstream bsid_out( ( out + ".bsid" ).c_str() );
ALL_SAMPLES.print( bsid_out );
bsid_out.close();
}
}
// mine(): main function for GERMLINE
void GERMLINE::mine( string params, string map, string ped,string outfile)
{
PolymorphicIndividualsExtractor * pie = inputManager.getPie();
inputManager.getIndividuals(map, ped);
if ( ! pie->valid() ) return;
string out;
if (outfile == "")
out = inputManager.getOutput();
else
out = outfile;
num_samples = 0;
num_matches = 0;
pie->loadInput();
MatchesBuilder mb( pie );
unsigned long long init_mem = (unsigned long long)(mem_all_matches+mem_bufferchr+mem_chromosome+mem_ind+mem_inds+mem_matchfactory+mem_markers+mem_snps+mem_window);
mem_expected_data= mb.calculateMemData();
if ( (init_mem+mem_expected_data) < MEM_BOUND)
{
cerr<<"\n\n\tBeginning Analysis"<<endl;
ofstream fout( ( out + ".log" ).c_str() );
fout << setw(65) << setfill('-') << ' ' << endl << setfill(' ');
fout << " Welcome to GERMLINE, a tool for detecting long segments shared" << endl;
fout << " by descent between pairs of individuals in near-linear time." << endl;
fout << endl;
fout << " For more details, please see the paper [ PMID: 18971310 ]" << endl;
fout << " or the web-site [ http://www.cs.columbia.edu/~gusev/germline/ ]" << endl;
fout << endl;
fout << " GERMLINE was coded by Alexander Gusev and collaborators in " << endl;
fout << " Itsik Pe'er's Computational Biology Lab at Columbia University" << endl;
fout << setw(65) << setfill('-') << ' ' << endl << setfill(' ');
if ( BINARY_OUT ) MATCH_FILE.open( ( out + ".bmatch" ).c_str() , ios::binary );
else MATCH_FILE.open( ( out + ".match" ).c_str() );
fout << params << endl;
fout << setw(65) << setfill('-') << ' ' << endl << setfill(' ');
fout << setw(50) << left << "Minimum match length: " << MIN_MATCH_LEN << " cM" << endl;
fout << setw(50) << "Allowed mismatch: " << MAX_ERR_HOMp << "% " << MAX_ERR_HETp << "%"<<endl;
fout << setw(50) << "Minimum Word size: " << MIN_WINDOW_SIZE << endl;
if ( ROI )
fout << setw(50) << "Target region: " << ALL_SNPS.getROIStart().getSNPID() << " - " << ALL_SNPS.getROIEnd().getSNPID() << endl;
else
fout << setw(50) << "Target region: " << "all" << endl;
time_t timer[2]; time( &timer[0] );
if ( DEBUG ) cout << "DEBUG MODE ON" << endl;
if ( ROI )
{
ALL_SNPS.beginChromosome();
ALL_SNPS_CURRENT_SIZE = ALL_SNPS.currentSize();
mb.buildMatches();
if ( !SILENT ) cout << "\nMatches completed ... freeing memory" << endl;
ALL_SAMPLES.freeMatches();
ALL_SAMPLES.freeMarkers();
WINDOWS_LIST.clear();
}
else
{
for ( ALL_SNPS.beginChromosome() ; ALL_SNPS.moreChromosome() ; ALL_SNPS.nextChromosome() )
{
MAX_WINDOW_SIZE=0;
ALL_SNPS_CURRENT_SIZE = ALL_SNPS.currentSize();
mb.buildMatches();
if ( !SILENT ) cout << "\nMatches completed ... freeing memory" << endl;
ALL_SAMPLES.freeMatches();
ALL_SAMPLES.freeMarkers();
WINDOWS_LIST.clear(); LAST_SET=false;
}
}
time( &timer[1] );
fout << setw(50) << "Total IBD segments: " << num_matches << endl;
fout << setw(50) << "Total runtime (sec): " << difftime( timer[1] , timer[0] ) << endl;
fout.close();
MATCH_FILE.close();
if ( BINARY_OUT )
{
ofstream bmid_out( ( out + ".bmid" ).c_str() );
ALL_SNPS.print( bmid_out );
bmid_out.close();
ofstream bsid_out( ( out + ".bsid" ).c_str() );
ALL_SAMPLES.print( bsid_out );
bsid_out.close();
}}
else
{
cerr<<"\n\n\tNot enough memory to load data...Cannot begin analysis";
cerr<<"\n\tRequire atleast "<<ceil((float)(init_mem+mem_expected_data)/1048576)<<" Mb space"<<endl;
}
}