static int load_discordant_reads(MEI_data& mei_data, std::vector<bam_info>& bam_sources, const std::string& chr_name,
const SearchWindow& window, UserDefinedSettings* userSettings) {
// Loop over associated bam files.
for (size_t i = 0; i < bam_sources.size(); i++) {
// Locate file.
bam_info source = bam_sources.at(i);
LOG_DEBUG(*logStream << time_log() << "Loading discordant reads from " << source.BamFile << std::endl);
// Setup link to bamfile, its index and header.
bamFile fp = bam_open(source.BamFile.c_str(), "r");
bam_index_t *idx = bam_index_load(source.BamFile.c_str());
if (idx == NULL) {
LOG_WARN(*logStream << time_log() << "Failed to load index for " << source.BamFile.c_str() << std::endl);
LOG_WARN(*logStream << "Skipping window: " << chr_name << ", " << window.getStart() << "--" <<
window.getEnd() << " for BAM-file: " << source.BamFile.c_str() << std::endl);
continue;
}
bam_header_t *header = bam_header_read(fp);
bam_init_header_hash(header);
int tid = bam_get_tid(header, chr_name.c_str());
if (tid < 0) {
LOG_WARN(*logStream << time_log() << "Could not find sequence in alignment file: '" << chr_name <<
"'" << std::endl);
LOG_WARN(*logStream << "Skipping window: " << chr_name << ", " << window.getStart() << "--" <<
window.getEnd() << " for BAM-file: " << source.BamFile.c_str() << std::endl);
continue;
}
mei_data.sample_names = get_sample_dictionary(header);
// Save insert size of current bamfile in data object provided for callback function.
// Note: the insert size should ideally be separate from the MEI_data object, tried to do
// this using a std::pair object, which did not work. Suggestions are welcome here.
mei_data.current_insert_size = source.InsertSize;
mei_data.current_chr_name = chr_name;
// Set up environment variable for callback function.
std::pair<MEI_data*, UserDefinedSettings*> env = std::make_pair(&mei_data, userSettings);
// Load discordant reads into mei_data.
bam_fetch(fp, idx, tid, window.getStart(), window.getEnd(), &env, fetch_disc_read_callback);
bam_index_destroy(idx);
}
return 0;
}
int searchIndels(ControlState& currentState, unsigned NumBoxes, const SearchWindow& window )
{
static int Count_DI = 0;
static int Count_DI_Plus = 0;
static int Count_DI_Minus = 0;
unsigned CloseIndex, FarIndex;
std::vector<unsigned> DI[NumBoxes];
unsigned TempBoxIndex;
LOG_INFO(*logStream << "Searching deletion-insertions ... " << std::endl);
//UserDefinedSettings *userSettings = UserDefinedSettings::Instance();
for (unsigned ReadIndex = 0; ReadIndex < currentState.Reads_SR.size(); ReadIndex++) {
SPLIT_READ& currentRead = currentState.Reads_SR[ReadIndex];
if (currentRead.Used
|| currentRead.UP_Far.empty() || currentRead.FragName != currentRead.FarFragName) {
continue;
}
CloseIndex = currentRead.UP_Close.size() - 1;
FarIndex = currentRead.UP_Far.size() - 1;
if (currentRead.UP_Far[FarIndex].Mismatches + currentRead.UP_Close[CloseIndex].Mismatches
> (short) (1 + userSettings->Seq_Error_Rate * (currentRead.UP_Far[FarIndex].LengthStr + currentRead.UP_Close[CloseIndex].LengthStr))) {
continue;
}
if (currentRead.MatchedD == Plus) {
if (currentRead.UP_Far[FarIndex].Direction == Minus) {
if (currentRead.UP_Far[FarIndex].LengthStr + currentRead.UP_Close[CloseIndex].LengthStr < currentRead.getReadLength() &&
currentRead.UP_Far[FarIndex].LengthStr + currentRead.UP_Close[CloseIndex].LengthStr >= userSettings->Min_Num_Matched_Bases &&
currentRead.UP_Far[FarIndex].AbsLoc > currentRead.UP_Close[CloseIndex].AbsLoc + 1) {
currentRead.Left = currentRead.UP_Close[CloseIndex].AbsLoc - currentRead.UP_Close[CloseIndex].LengthStr + 1;
currentRead.Right = currentRead.UP_Far[FarIndex].AbsLoc + currentRead.UP_Far[FarIndex].LengthStr - 1;
currentRead.BP = currentRead.UP_Close[CloseIndex].LengthStr - 1;
currentRead.NT_size = currentRead.getReadLength() - currentRead.UP_Far[FarIndex].LengthStr - currentRead.UP_Close[CloseIndex].LengthStr;
currentRead.NT_str = ReverseComplement( currentRead. getUnmatchedSeq()). substr( currentRead.BP + 1, currentRead.NT_size);
currentRead.IndelSize = (currentRead.Right - currentRead.Left) + currentRead.NT_size - currentRead.getReadLengthMinus();
currentRead.BPLeft = currentRead.UP_Close[CloseIndex].AbsLoc - g_SpacerBeforeAfter;
currentRead.BPRight = currentRead.UP_Far[FarIndex].AbsLoc - g_SpacerBeforeAfter;
if (1) {
if (readTransgressesBinBoundaries( currentRead, window.getEnd())) {
saveReadForNextCycle(currentRead, currentState.FutureReads_SR);
} else {
if (readInSpecifiedRegion( currentRead, userSettings->getRegion() ) ) {
TempBoxIndex = (int) (currentRead. BPLeft) / BoxSize;
if (TempBoxIndex < NumBoxes) {
DI[TempBoxIndex]. push_back(ReadIndex);
currentRead.Used = true;
Count_DI++;
Count_DI_Plus++;
}
}
}
}
}
}
} else if (currentRead.MatchedD == Minus) {
if (currentRead.UP_Far[FarIndex].Direction == Plus) {
if (currentRead.UP_Close[CloseIndex].LengthStr + currentRead.UP_Far[FarIndex].LengthStr < currentRead.getReadLength() &&
currentRead.UP_Close[CloseIndex].LengthStr + currentRead.UP_Far[FarIndex].LengthStr >= userSettings->Min_Num_Matched_Bases &&
currentRead.UP_Close[CloseIndex].AbsLoc > currentRead.UP_Far[FarIndex].AbsLoc + 1) {
currentRead.Left = currentRead.UP_Far[FarIndex].AbsLoc - currentRead.UP_Far[FarIndex].LengthStr + 1;
currentRead.Right = currentRead.UP_Close[CloseIndex].AbsLoc + currentRead.UP_Close[CloseIndex].LengthStr - 1;
currentRead.BP = currentRead.UP_Far[FarIndex].LengthStr - 1;
currentRead.NT_size = currentRead.getReadLength() - currentRead.UP_Close[CloseIndex].LengthStr - currentRead.UP_Far[FarIndex].LengthStr;
currentRead.NT_str = currentRead.getUnmatchedSeq(). substr( currentRead.BP + 1, currentRead.NT_size);
currentRead.IndelSize = (currentRead.Right - currentRead.Left) - currentRead.getReadLengthMinus() + currentRead.NT_size;
currentRead.BPLeft = currentRead.UP_Far[FarIndex].AbsLoc - g_SpacerBeforeAfter;
currentRead.BPRight = currentRead.UP_Close[CloseIndex].AbsLoc - g_SpacerBeforeAfter;
{
if ( 1 ) {
if (readTransgressesBinBoundaries( currentRead, window.getEnd())) {
saveReadForNextCycle( currentRead, currentState.FutureReads_SR);
} else {
if (readInSpecifiedRegion( currentRead, userSettings->getRegion())) {
TempBoxIndex = (int) (currentRead. BPLeft) / BoxSize;
if (TempBoxIndex < NumBoxes) {
DI[TempBoxIndex]. push_back(ReadIndex);
currentRead.Used = true;
Count_DI++;
Count_DI_Minus++;
}
}
}
}
}
}
}
}
}
LOG_INFO(*logStream << "Total: " << Count_DI << "\t+" << Count_DI_Plus << "\t-"
<< Count_DI_Minus << std::endl);
std::ofstream DeletionOutf( userSettings->getDOutputFilename().c_str(), std::ios::app);
std::ofstream inversionsOutf( userSettings->getINVOutputFilename().c_str(), std::ios::app);
SortOutputDI(currentState, NumBoxes, window.getChromosome()->getSeq(), currentState.Reads_SR, DI, DeletionOutf, inversionsOutf);
DeletionOutf.close();
for (unsigned int i = 0; i < NumBoxes; i++) {
DI[i].clear();
}
return EXIT_SUCCESS;
}
int searchTandemDuplicationsNT(ControlState& currentState, unsigned NumBoxes, const SearchWindow& window)
{
static int Count_TD_NT = 0;
static int Count_TD_NT_Plus = 0;
static int Count_TD_NT_Minus = 0;
std::vector<unsigned> TD_NT[NumBoxes];
int CloseIndex = 0;
int FarIndex = 0;
//UserDefinedSettings* userSettings = UserDefinedSettings::Instance();
LOG_INFO(*logStream << "Searching tandem duplication events with non-template sequence ... " << std::endl);
for (unsigned ReadIndex = 0; ReadIndex < currentState.Reads_SR.size(); ReadIndex++) {
SPLIT_READ& currentRead = currentState.Reads_SR[ReadIndex];
if (currentRead.Used || currentRead.UP_Far.empty() || currentRead.FragName != currentRead.FarFragName) {
continue;
}
CloseIndex = currentRead.UP_Close.size() - 1;
FarIndex = currentRead.UP_Far.size() - 1;
if (currentRead.UP_Far[FarIndex].LengthStr + currentRead.UP_Close[CloseIndex].LengthStr >= currentRead.getReadLength()) {
continue;
}
if (currentRead.UP_Far[FarIndex].Mismatches + currentRead.UP_Close[CloseIndex].Mismatches >
(short) (1 + userSettings->Seq_Error_Rate * (currentRead.UP_Far[FarIndex].LengthStr + currentRead.UP_Close[CloseIndex].LengthStr))) {
continue;
}
if (currentRead.MatchedD == Plus) {
if (currentRead.UP_Far[FarIndex].Direction == Minus) {
if (currentRead.UP_Far[FarIndex].AbsLoc + currentRead.UP_Far[FarIndex].LengthStr < currentRead.UP_Close[CloseIndex].AbsLoc &&
currentRead.UP_Far[FarIndex].AbsLoc + currentRead.UP_Close[CloseIndex].LengthStr < currentRead.UP_Close[CloseIndex].AbsLoc &&
currentRead.UP_Far[FarIndex].LengthStr + currentRead.UP_Close[CloseIndex].LengthStr > userSettings->Min_Num_Matched_Bases) {
currentRead.Right = currentRead.UP_Close[CloseIndex].AbsLoc - currentRead.UP_Close[CloseIndex].LengthStr + 1;
currentRead.Left = currentRead.UP_Far[FarIndex].AbsLoc + currentRead.UP_Far[FarIndex].LengthStr - 1;
currentRead.BP = currentRead.UP_Close[CloseIndex].LengthStr - 1;
currentRead.IndelSize = currentRead.UP_Close[CloseIndex].AbsLoc - currentRead.UP_Far[FarIndex].AbsLoc + 1;
currentRead.NT_size = currentRead.getReadLength() - currentRead.UP_Close[CloseIndex].LengthStr - currentRead.UP_Far[FarIndex].LengthStr;
currentRead.NT_str = currentRead.getUnmatchedSeqRev().substr( currentRead.BP + 1, currentRead.NT_size);
currentRead.BPRight = currentRead.UP_Close[CloseIndex].AbsLoc - g_SpacerBeforeAfter;
currentRead.BPLeft = currentRead.UP_Far[FarIndex].AbsLoc - g_SpacerBeforeAfter;
if (readTransgressesBinBoundaries( currentRead, window.getEnd())) {
saveReadForNextCycle(currentRead, currentState.FutureReads_SR);
}
else {
if ( readInSpecifiedRegion( currentRead, userSettings->getRegion())) {
TD_NT[(int) currentRead. BPLeft / BoxSize]. push_back(ReadIndex);
currentRead.Used = true;
Count_TD_NT++;
Count_TD_NT_Plus++;
}
}
}
}
}
else if (currentRead.MatchedD == Minus) {
if (currentRead.UP_Far[FarIndex].Direction == Plus) {
if (currentRead.UP_Close[CloseIndex].AbsLoc + currentRead.UP_Close[CloseIndex].LengthStr < currentRead.UP_Far[FarIndex].AbsLoc &&
currentRead.UP_Close[CloseIndex].AbsLoc + currentRead.UP_Far[FarIndex].LengthStr < currentRead.UP_Far[FarIndex].AbsLoc &&
currentRead.UP_Far[FarIndex].LengthStr + currentRead.UP_Close[CloseIndex].LengthStr > userSettings->Min_Num_Matched_Bases) {
currentRead.Right = currentRead.UP_Far[FarIndex].AbsLoc - currentRead.UP_Far[FarIndex].LengthStr + 1;
currentRead.Left = currentRead.UP_Close[CloseIndex].AbsLoc + currentRead.UP_Close[CloseIndex].LengthStr - 1;
currentRead.BP = currentRead.UP_Far[FarIndex].LengthStr - 1;
currentRead.IndelSize = currentRead.UP_Far[FarIndex].AbsLoc - currentRead.UP_Close[CloseIndex].AbsLoc + 1;
currentRead.NT_size = currentRead.getReadLength() - currentRead.UP_Close[CloseIndex].LengthStr - currentRead.UP_Far[FarIndex].LengthStr;
currentRead.NT_str = currentRead.getUnmatchedSeq(). substr( currentRead.BP + 1, currentRead.NT_size);
currentRead.BPRight = currentRead.UP_Far[FarIndex].AbsLoc - g_SpacerBeforeAfter;
currentRead.BPLeft = currentRead.UP_Close[CloseIndex].AbsLoc - g_SpacerBeforeAfter;
if (readTransgressesBinBoundaries( currentRead, window.getEnd())) {
saveReadForNextCycle(currentRead, currentState.FutureReads_SR);
}
else {
if ( readInSpecifiedRegion( currentRead, userSettings->getRegion())) {
TD_NT[(int) currentRead. BPLeft / BoxSize]. push_back(ReadIndex);
currentRead.Used = true;
Count_TD_NT++;
Count_TD_NT_Minus++;
}
}
}
}
}
}
LOG_INFO(*logStream << "Total: " << Count_TD_NT << "\t+" << Count_TD_NT_Plus << "\t-" << Count_TD_NT_Minus << std::endl);
std::ofstream TDOutf(userSettings->getTDOutputFilename().c_str(), std::ios::app);
SortAndOutputTandemDuplications(currentState, NumBoxes, window.getChromosome()->getSeq(), currentState.Reads_SR, TD_NT, TDOutf, true);
TDOutf.close();
for (unsigned int i = 0; i < NumBoxes; i++) {
TD_NT[i].clear();
}
return EXIT_SUCCESS;
}
// std::cout << currentRead.BPLeft << " " << BoxSize << " " << NumBoxes << " " << TempBoxIndex << std::endl;
int SearchVariant::Search(BDData & g_bdData, ControlState& currentState, const unsigned NumBoxes, const SearchWindow& window)
{
std::vector<unsigned> Vars[NumBoxes];
unsigned TempBoxIndex;
unsigned int farEndExists = 0;
unsigned int readsUsed = 0;
unsigned int bpSum = 0;
for (unsigned ReadIndex = 0; ReadIndex < currentState.Reads_SR.size(); ReadIndex++) {
SPLIT_READ& currentRead = currentState.Reads_SR[ReadIndex];
if (currentRead.Used) {
readsUsed++;
}
if (!currentRead.UP_Far.empty()) {
farEndExists++;
bpSum += currentRead.UP_Far[ currentRead.UP_Far.size()-1 ].AbsLoc;
}
if (bpSum>1000000000) {
bpSum -= 1000000000;
}
}
*logStream << "Reads already used: " << readsUsed << std::endl;
*logStream << "Far ends already mapped " << farEndExists << std::endl;
*logStream << "Checksum of far ends: " << bpSum << std::endl;
//UserDefinedSettings* userSettings = UserDefinedSettings::Instance();
LOG_INFO(*logStream << "Searching " << typeOfVariant << " ... " << std::endl);
const std::string ChrSeq = window.getChromosome()->getSeq();
for (unsigned ReadIndex = 0; ReadIndex < currentState.Reads_SR.size(); ReadIndex++) {
SPLIT_READ& currentRead = currentState.Reads_SR[ReadIndex];
//std::cout << ReadIndex << std::endl;
//std::cout << currentRead << std::endl;
if (currentRead.FragName != currentRead.FarFragName) {
//std::cout << "inter-chromosome split-read" << std::endl;
continue;
}
if (currentRead.Used || currentRead.UP_Far.empty()) {
continue;
}
if (currentRead.MatchedD == Plus) {
for (short MAX_SNP_ERROR_index = 0; MAX_SNP_ERROR_index <= currentRead.getMAX_SNP_ERROR(); MAX_SNP_ERROR_index++) {
if (currentRead.Used) {
break;
}
for (unsigned int CloseIndex = 0; CloseIndex
< currentRead.UP_Close.size(); CloseIndex++) {
if (currentRead.Used) {
break;
}
if (currentRead.UP_Close[CloseIndex].Mismatches > MAX_SNP_ERROR_index) {
continue;
}
for (int FarIndex = currentRead.UP_Far.size() - 1; FarIndex >= 0; FarIndex--) {
if (currentRead.Used) {
break;
}
if (currentRead.UP_Far[FarIndex]. Mismatches > MAX_SNP_ERROR_index) {
continue;
}
if (currentRead.UP_Far[FarIndex]. Mismatches + currentRead.UP_Close[CloseIndex]. Mismatches > MAX_SNP_ERROR_index) {
continue;
}
if (currentRead.UP_Far[FarIndex]. Direction == Minus) {
if (decisionBranch1(currentRead, CloseIndex, FarIndex)) {
currentRead.Left = currentRead. UP_Close[CloseIndex].AbsLoc - currentRead. UP_Close[CloseIndex].LengthStr + 1;
currentRead.Right = currentRead.UP_Far[FarIndex]. AbsLoc + currentRead.UP_Far[FarIndex]. LengthStr - 1;
currentRead.BP = currentRead. UP_Close[CloseIndex].LengthStr - 1;
currentRead.IndelSize = calculateIndelSize(currentRead);
currentRead.NT_str = getInsertedStr1(currentRead);
currentRead.BPLeft = currentRead. UP_Close[CloseIndex].AbsLoc - g_SpacerBeforeAfter;
currentRead.BPRight = currentRead.UP_Far[FarIndex]. AbsLoc - g_SpacerBeforeAfter;
unsigned RealBP_left= currentRead.BPLeft;
unsigned RealBP_right = currentRead.BPRight;//, DIFF;
if (ChrSeq.size() < RealBP_left || ChrSeq.size() < RealBP_right) {
currentRead.Used = true;
break;
}
if (currentRead.NT_str.size()) {
//std::cout << "ns i currentRead.NT_str.size()" << std::endl;
//std::cout << currentRead.NT_str << " " << RealBP_left << " " << RealBP_right << std::endl;
GetRealStart4Insertion(ChrSeq, currentRead.NT_str, RealBP_left, RealBP_right);
//std::cout << "ne i currentRead.NT_str.size()" << std::endl;
} else {
//std::cout << "no_ns d currentRead.NT_str.size()" << std::endl;
//std::cout << RealBP_left << " " << RealBP_right << std::endl;
GetRealStart4Deletion(ChrSeq, RealBP_left, RealBP_right);
//std::cout << "no_ns d currentRead.NT_str.size()" << std::endl;
}
short DIFF = currentRead.BPLeft - RealBP_left;
DIFF = !((currentRead.BP - 1)<DIFF)?DIFF:(currentRead.BP - 1); // min(DIFF, currentRead.BP - 1);
if (DIFF > 0) {
//std::cout << "DIFF " << DIFF << std::endl;
currentRead.BP -= DIFF;
currentRead.BPLeft -= DIFF;
currentRead.BPRight -= DIFF;
}
if (readTransgressesBinBoundaries( currentRead, window.getEnd())) {
saveReadForNextCycle( currentRead, currentState.FutureReads_SR);
} else {
if (readInSpecifiedRegion( currentRead, userSettings->getRegion())) {
TempBoxIndex = (int) (currentRead. BPLeft) / BoxSize;
//std::cout << currentRead.BPLeft << " " << BoxSize << " " << NumBoxes << " " << TempBoxIndex << std::endl;
if (TempBoxIndex < NumBoxes) {
Vars[TempBoxIndex].push_back(ReadIndex);
currentRead.Used = true;
Count_Var_Plus++;
Count_Var++;
}
}
}
}
}
}
}
}
} else if (currentRead.MatchedD == Minus) {
for (short MAX_SNP_ERROR_index = 0; MAX_SNP_ERROR_index <= currentRead.getMAX_SNP_ERROR(); MAX_SNP_ERROR_index++) {
if (currentRead.Used) {
break;
}
for (int CloseIndex = currentRead.UP_Close.size() - 1; CloseIndex >= 0; CloseIndex--) {
if (currentRead.Used) {
break;
}
if (currentRead.UP_Close[CloseIndex]. Mismatches > MAX_SNP_ERROR_index) {
continue;
}
for (int FarIndex = currentRead.UP_Far.size() - 1; FarIndex >= 0; FarIndex--) {
if (currentRead.Used) {
break;
}
if (currentRead.UP_Far[FarIndex]. Mismatches > MAX_SNP_ERROR_index) {
continue;
}
if (currentRead.UP_Far[FarIndex]. Mismatches + currentRead.UP_Close[CloseIndex]. Mismatches > MAX_SNP_ERROR_index) {
continue;
}
if (currentRead.UP_Far[FarIndex]. Direction == Plus) {
if (decisionBranch2(currentRead, CloseIndex, FarIndex)) {
currentRead.Left = currentRead.UP_Far[FarIndex]. AbsLoc - currentRead.UP_Far[FarIndex]. LengthStr + 1;
currentRead.Right = currentRead. UP_Close[CloseIndex].AbsLoc + currentRead. UP_Close[CloseIndex].LengthStr - 1;
currentRead.BP = currentRead.UP_Far[FarIndex]. LengthStr - 1;
currentRead.IndelSize = calculateIndelSize(currentRead);
currentRead.NT_str = getInsertedStr2(currentRead);
currentRead.BPLeft = currentRead.UP_Far[FarIndex].AbsLoc - g_SpacerBeforeAfter;
currentRead.BPRight = currentRead.UP_Close[CloseIndex].AbsLoc - g_SpacerBeforeAfter;
unsigned RealBP_left= currentRead.BPLeft;
unsigned RealBP_right = currentRead.BPRight;//, DIFF;
if (ChrSeq.size() < RealBP_left || ChrSeq.size() < RealBP_right) {
currentRead.Used = true;
break;
}
if (currentRead.NT_str.size()) {
//std::cout << "ns i currentRead.NT_str.size()" << std::endl;
//std::cout << currentRead.NT_str << " " << RealBP_left << " " << RealBP_right << std::endl;
GetRealStart4Insertion(ChrSeq, currentRead.NT_str, RealBP_left, RealBP_right);
//std::cout << "ne i currentRead.NT_str.size()" << std::endl;
} else {
//std::cout << "no_ns d currentRead.NT_str.size()" << std::endl;
//std::cout << RealBP_left << " " << RealBP_right << std::endl;
GetRealStart4Deletion(ChrSeq, RealBP_left, RealBP_right);
//std::cout << "no_ne d currentRead.NT_str.size()" << std::endl;
}
short DIFF = currentRead.BPLeft - RealBP_left;
DIFF = !((currentRead.BP - 1)<DIFF)?DIFF:(currentRead.BP - 1);
if (DIFF > 0) {
// std::cout << DIFF << std::endl;
//std::cout << "DIFF " << DIFF << std::endl;
currentRead.BP -= DIFF;
currentRead.BPLeft -= DIFF;
currentRead.BPRight -= DIFF;
}
if (readTransgressesBinBoundaries( currentRead, window.getEnd())) {
saveReadForNextCycle( currentRead, currentState.FutureReads_SR);
} else {
if (readInSpecifiedRegion( currentRead, userSettings->getRegion())) {
TempBoxIndex = (int) (currentRead. BPLeft) / BoxSize;
//std::cout << currentRead.BPLeft << " " << BoxSize << " " << NumBoxes << " " << TempBoxIndex << std::endl;
if (TempBoxIndex < NumBoxes) {
Vars[TempBoxIndex]. push_back(ReadIndex);
currentRead.Used = true;
Count_Var++;
Count_Var_Minus++;
}
}
}
}
}
}
}
}
}
}
LOG_INFO(*logStream << "Total: " << Count_Var << "\t+" << Count_Var_Plus << "\t-" << Count_Var_Minus << std::endl);
outputResults(g_bdData, currentState, Vars, NumBoxes, window);
for (unsigned int i = 0; i < NumBoxes; i++) {
Vars[i].clear();
}
return EXIT_SUCCESS;
}
int searchTandemDuplications(ControlState& currentState, unsigned NumBoxes, const SearchWindow& window)
{
static int Count_TD = 0;
static int Count_TD_Plus = 0;
static int Count_TD_Minus = 0;
std::vector<unsigned> TD[NumBoxes];
//UserDefinedSettings* userSettings = UserDefinedSettings::Instance();
unsigned TempBoxIndex;
LOG_INFO(*logStream << "Searching tandem duplication events ... " << std::endl);
//std::cout << "std1" << std::endl;
for (unsigned ReadIndex = 0; ReadIndex < currentState.Reads_SR.size(); ReadIndex++) {
SPLIT_READ& currentRead = currentState.Reads_SR[ReadIndex];
//std::cout << "std1a" << std::endl;
if (currentRead.Used || currentRead.UP_Far.empty() || currentRead.FragName != currentRead.FarFragName) {
continue;
}
//std::cout << "std1b" << std::endl;
if (currentRead.MatchedD == Plus) {
//std::cout << "in TD plus" << std::endl;
for (short MAX_SNP_ERROR_index = 0; MAX_SNP_ERROR_index <= currentRead.getMAX_SNP_ERROR(); MAX_SNP_ERROR_index++) {
//std::cout << "std1c" << std::endl;
for (unsigned int CloseIndex = 0; CloseIndex < currentRead.UP_Close.size(); CloseIndex++) {
//std::cout << "std1d" << std::endl;
if (currentRead.Used /* || currentRead. BPLeft == 0*/) {
break;
}
if (currentRead.UP_Close[CloseIndex].Mismatches > MAX_SNP_ERROR_index) {
continue;
}
//std::cout << "std1e " << currentRead.UP_Far.size() << std::endl;
for (int FarIndex = currentRead.UP_Far.size() - 1; FarIndex >= 0; FarIndex--) {
//std::cout << "std1f" << std::endl;
if (currentRead.Used) {
break;
}
if (currentRead.UP_Far[FarIndex].Mismatches > MAX_SNP_ERROR_index) {
continue;
}
if (currentRead.UP_Far[FarIndex].Mismatches + currentRead.UP_Close[CloseIndex].Mismatches > MAX_SNP_ERROR_index) {
continue;
}
//std::cout << "std1g" << std::endl;
if (currentRead.UP_Far[FarIndex].Direction == Minus) {
if (currentRead.UP_Far[FarIndex].LengthStr + currentRead.UP_Close[CloseIndex].LengthStr == currentRead.getReadLength() &&
currentRead.UP_Far[FarIndex].AbsLoc + currentRead.UP_Far[FarIndex].LengthStr < currentRead.UP_Close[CloseIndex].AbsLoc &&
currentRead.UP_Far[FarIndex].AbsLoc + currentRead.UP_Close[CloseIndex].LengthStr < currentRead.UP_Close[CloseIndex].AbsLoc) {
currentRead.Right = currentRead.UP_Close[CloseIndex].AbsLoc - currentRead.UP_Close[CloseIndex].LengthStr + 1;
currentRead.Left = currentRead. UP_Far[FarIndex].AbsLoc + currentRead. UP_Far[FarIndex].LengthStr - 1;
currentRead.BP = currentRead.UP_Close[CloseIndex].LengthStr - 1;
currentRead.IndelSize = currentRead.UP_Close[CloseIndex].AbsLoc - currentRead. UP_Far[FarIndex].AbsLoc + 1;
currentRead.BPRight = currentRead.UP_Close[CloseIndex].AbsLoc - g_SpacerBeforeAfter;
currentRead.BPLeft = currentRead.UP_Far[FarIndex].AbsLoc - g_SpacerBeforeAfter;
if (currentRead. BPLeft == 0) {
continue;
}
LeftMostTD(currentState, currentRead, window);
if (readTransgressesBinBoundaries( currentRead, window.getEnd())) {
saveReadForNextCycle( currentRead, currentState.FutureReads_SR);
} else {
if (readInSpecifiedRegion( currentRead, userSettings->getRegion())) {
TempBoxIndex = (int) (currentRead. BPLeft) / BoxSize;
if (TempBoxIndex < NumBoxes) {
//std::cout << currentRead << std::endl;
TD[TempBoxIndex]. push_back(ReadIndex);
currentRead.Used = true;
Count_TD++;
Count_TD_Plus++;
}
}
}
}
}
}
}
}
} else if (currentRead.MatchedD == Minus) {
//std::cout << "in Minus" << std::endl;
for (short MAX_SNP_ERROR_index = 0; MAX_SNP_ERROR_index <= currentRead.getMAX_SNP_ERROR(); MAX_SNP_ERROR_index++) {
//std::cout << "in Minus 1" << std::endl;
for (int CloseIndex = currentRead.UP_Close.size() - 1; CloseIndex >= 0; CloseIndex--) {
//std::cout << "in Minus 2" << std::endl;
if (currentRead.Used /*|| currentRead. BPLeft == 0*/) {
break;
}
if (currentRead.UP_Close[CloseIndex].Mismatches > MAX_SNP_ERROR_index) {
continue;
}
//std::cout << "in Minus 3" << std::endl;
for (int FarIndex = 0; FarIndex < (int) currentRead.UP_Far.size(); FarIndex++) {
if (currentRead.Used) {
break;
}
if (currentRead.UP_Far[FarIndex].Mismatches > MAX_SNP_ERROR_index) {
continue;
}
if (currentRead.UP_Far[FarIndex].Mismatches + currentRead.UP_Close[CloseIndex].Mismatches > MAX_SNP_ERROR_index) {
continue;
}
if (currentRead.UP_Far[FarIndex]. Direction == Plus) {
//std::cout << "BPLeft" << currentRead. BPLeft << std::endl;
if (currentRead.UP_Close[CloseIndex].LengthStr + currentRead.UP_Far[FarIndex].LengthStr == currentRead.getReadLength() &&
currentRead.UP_Close[CloseIndex].AbsLoc + currentRead.UP_Close[CloseIndex].LengthStr < currentRead.UP_Far[FarIndex]. AbsLoc &&
currentRead.UP_Close[CloseIndex].AbsLoc + currentRead.UP_Far[FarIndex].LengthStr < currentRead.UP_Far[FarIndex]. AbsLoc) {
currentRead.Right = currentRead. UP_Far[FarIndex].AbsLoc - currentRead. UP_Far[FarIndex].LengthStr + 1;
currentRead.Left = currentRead.UP_Close[CloseIndex].AbsLoc + currentRead.UP_Close[CloseIndex].LengthStr - 1;
currentRead.BP = currentRead. UP_Far[FarIndex].LengthStr - 1;
currentRead.IndelSize = currentRead. UP_Far[FarIndex].AbsLoc - currentRead.UP_Close[CloseIndex].AbsLoc + 1;
currentRead.BPRight = currentRead.UP_Far[FarIndex].AbsLoc - g_SpacerBeforeAfter;
currentRead.BPLeft = currentRead.UP_Close[CloseIndex].AbsLoc - g_SpacerBeforeAfter;
//std::cout << "in Minus 4a" << std::endl;
if (currentRead. BPLeft == 0) {
continue;
}
LeftMostTD(currentState, currentRead, window);
//std::cout << "in Minus 4b" << std::endl;
if (readTransgressesBinBoundaries( currentRead, window.getEnd())) {
saveReadForNextCycle( currentRead, currentState.FutureReads_SR);
} else {
if (readInSpecifiedRegion( currentRead, userSettings->getRegion())) {
TempBoxIndex = (int) (currentRead. BPLeft) / BoxSize;
if (TempBoxIndex < NumBoxes) {
//std::cout << currentRead << std::endl;
TD[TempBoxIndex]. push_back(ReadIndex);
currentRead.Used = true;
Count_TD++;
Count_TD_Minus++;
}
}
}
}
}
}
//std::cout << "in Minus 4" << std::endl;
}
}
}
}
//std::cout << "std2" << std::endl;
LOG_INFO(*logStream << "Total: " << Count_TD << "\t+" << Count_TD_Plus << "\t-" << Count_TD_Minus << std::endl);
//std::cout << "TD 1" << std::endl;
std::ofstream TDOutf(userSettings->getTDOutputFilename().c_str(), std::ios::app);
SortAndOutputTandemDuplications(currentState, NumBoxes, window.getChromosome()->getSeq(), currentState.Reads_SR, TD, TDOutf, false);
//std::cout << "TD 2" << std::endl;
//std::cout << "std3" << std::endl;
for (unsigned int i = 0; i < NumBoxes; i++) {
TD[i].clear();
}
// std::cout << "std4" << std::endl;
return EXIT_SUCCESS;
}