uint Aligner::checkBeginGreedy(const string& read, pair<kmer, uint>& overlap, vector<uNumber>& path, uint errors){
if(overlap.second==0){path.push_back(0);return 0;}
string readLeft(read.substr(0,overlap.second)),unitig;
auto rangeUnitigs(getEnd(overlap.first));
uint minMiss(errors+1),indiceMinMiss(0);
bool ended(false);
int offset(0);
kmer nextOverlap(0);
for(uint i(0); i<rangeUnitigs.size(); ++i){
unitig=(rangeUnitigs[i].first);
if(unitig.size()-k+1>=readLeft.size()){
uint miss(missmatchNumber(unitig.substr(unitig.size()-readLeft.size()-k+1,readLeft.size()),readLeft, errors));
// if(miss==0){
// path.push_back(unitig.size()-readLeft.size()-k+1);
// path.push_back(rangeUnitigs[i].second);
// return 0;
// }
if(miss<minMiss){
minMiss=miss;
indiceMinMiss=i;
ended=true;
offset=unitig.size()-readLeft.size()-k+1;
}
}else{
uint miss(missmatchNumber(unitig.substr(0,unitig.size()-k+1), readLeft.substr(readLeft.size()+k-1-unitig.size()), errors));
// if(miss==0){
// minMiss+=mapOnLeftEndGreedy(read, path, {nextOverlap,overlap.second-(nextUnitig.size()-k+1)},errors);
// if(minMiss<=errors){
// path.push_back(rangeUnitigs[indiceMinMiss].second);
// sucessML++;
// }
// }
if(miss<minMiss){
kmer overlapNum(str2num(unitig.substr(0,k-1)));
if(miss<minMiss){
ended=false;
minMiss=miss;
indiceMinMiss=i;
nextOverlap=overlapNum;
}
}
}
}
if(minMiss<=errors){
if(ended){
path.push_back(offset);
path.push_back(rangeUnitigs[indiceMinMiss].second);
return minMiss;
}
minMiss+=mapOnLeftEndGreedy(read, path, {nextOverlap,overlap.second-(rangeUnitigs[indiceMinMiss].first.size()-k+1)},errors-minMiss);
if(minMiss<=errors){
path.push_back(rangeUnitigs[indiceMinMiss].second);
sucessML++;
}
}
return minMiss;
}
uint Aligner::mapOnRightEndGreedy(const string &read, vector<uNumber>& path, const pair<kmer, uint>& overlap , uint errors){
// cout<<"moreg"<<endl;
string unitig,readLeft(read.substr(overlap.second)),nextUnitig;
// auto rangeUnitigs(getBeginOpti(overlap.first,path.back()));
auto rangeUnitigs(getBegin(overlap.first));
uint miniMiss(errors+1), miniMissIndice(9);
bool ended(false);
// int offset(0);
kmer nextOverlap(0);
// cout<<"go"<<endl;
// for(uint i(0); i<rangeUnitigs2.size(); ++i){
// cout<<(rangeUnitigs2[i].first)<<endl;
// }
// cout<<"true"<<endl;
for(uint i(0); i<rangeUnitigs.size(); ++i){
unitig=(rangeUnitigs[i].first);
// bool stop(false);
// if(rangeUnitigs[i].first!=rangeUnitigs2[i].first){
// cout<<read<<endl;
// cout<<"lol2"<<endl;
// cout<<rangeUnitigs[i].first<<endl<<rangeUnitigs2[i].first<<endl;
// cout<<rangeUnitigs[i].second<<" "<<rangeUnitigs2[i].second<<endl;
// stop=true;
// }
// if(stop){cin.get();}
// cout<<unitig<<endl;
// cout<<rangeUnitigs[i].first<<endl;
//case the rest of the read is too small
if(readLeft.size()<=unitig.size()){
uint miss(missmatchNumber(unitig.substr(0,readLeft.size()), readLeft, errors));
if(miss<miniMiss){
miniMiss=miss;
miniMissIndice=i;
ended=true;
// offset=unitig.size()-readLeft.size()-k+1;
}
}else{
//case the read is big enough we want to recover a true overlap
uint miss(missmatchNumber(unitig, read.substr(overlap.second,unitig.size()), errors));
if(miss<miniMiss){
if(miss<miniMiss){
kmer overlapNum(str2num(unitig.substr(unitig.size()-k+1,k-1)));
miniMiss=miss;
miniMissIndice=i;
nextUnitig=unitig;
nextOverlap=overlapNum;
}
}
}
}
// cout<<"end"<<endl;
if(miniMiss<=errors){
path.push_back(rangeUnitigs[miniMissIndice].second);
if (ended){return miniMiss;}
miniMiss+=mapOnRightEndGreedy(read , path, {nextOverlap,overlap.second+(nextUnitig.size()-k+1)}, errors-miniMiss);
}
return miniMiss;
}
uint Aligner::checkEndExhaustive(const string& read, pair<kmer, uint>& overlap, vector<uNumber>& path, uint errors){
string readLeft(read.substr(overlap.second+k-1)),unitig;
vector<uNumber> path2keep;
if(readLeft.empty()){
path.push_back(0);
return 0;
}
auto rangeUnitigs(getBegin(overlap.first));
uint minMiss(errors+1),indiceMinMiss(9);
bool ended(false);
int offset(-2);
if(partial & rangeUnitigs.empty()){
//if(!path.empty()){
return 0;
//}
}
for(uint i(0); i<rangeUnitigs.size(); ++i){
unitig=(rangeUnitigs[i].first);
if(unitig.size()-k+1>=readLeft.size()){
uint miss(missmatchNumber(unitig.substr(k-1,readLeft.size()),readLeft, errors));
if(miss<minMiss){
minMiss=miss;
indiceMinMiss=i;
ended=true;
offset=readLeft.size()+k-1;
}
}else{
uint miss(missmatchNumber(unitig.substr(k-1),readLeft.substr(0,unitig.size()-k+1), errors));
if(miss<minMiss){
kmer overlapNum(str2num(unitig.substr(unitig.size()-k+1,k-1)));
vector<uNumber> possiblePath;
miss+=mapOnRightEndExhaustive(read, possiblePath, {overlapNum,overlap.second+(unitig.size()-k+1)},errors-miss);
if(miss<minMiss){
path2keep=possiblePath;
minMiss=miss;
indiceMinMiss=i;
ended=false;
}
}
}
}
if(minMiss<=errors){
if(ended){
path.push_back(rangeUnitigs[indiceMinMiss].second);
path.push_back(offset);
}else{
path.push_back(rangeUnitigs[indiceMinMiss].second);
path.insert(path.end(), path2keep.begin(),path2keep.end());
}
}
return minMiss;
}
pair<uint,uint> Aligner::mapOnRightCache(const string &read, vector<uNumber>& path, const overlapStruct& overlap, const vector<overlapStruct>& listOverlap, bool& ended,uint start, uint errors){
string unitig, readLeft(read.substr(overlap.pos+k-1)),nextUnitig;
if(readLeft.empty()){cout<<"should not appears"<<endl;exit(0);return {start,0};}
auto rangeUnitigs(overlap.unitig);
uint miniMiss(errors+1),miniMissIndice(9);
uint next(start);
kmer nextOverlapNum(0);
for(uint i(0); i<rangeUnitigs.size(); ++i){
unitig=(rangeUnitigs[i]);
//case the rest of the read is too small
if(readLeft.size() <= unitig.size()-k+1){
uint miss(missmatchNumber(unitig.substr(k-1,readLeft.size()), readLeft, errors));
if(miss<miniMiss){
ended=true;
miniMiss=miss;
miniMissIndice=i;
}
}else{
//case the read is big enough we want to recover a true overlap
uint miss(missmatchNumber(unitig.substr(k-1), readLeft.substr(0,unitig.size()-k+1), errors));
if(miss<miniMiss){
kmer overlapNum(str2num(unitig.substr(unitig.size()-k+1,k-1)));
if(miss<miniMiss){
ended=false;
miniMiss=miss;
miniMissIndice=i;
nextOverlapNum=overlapNum;
nextUnitig=unitig;
next=start;
for(uint j(start+1); j<listOverlap.size(); ++j){
if(overlapNum==listOverlap[j].seq and listOverlap[j].pos==overlap.pos+unitig.size()-k+1){
next=j;
}
}
}
}
}
}
if(ended){
path.push_back(overlap.unitigNumbers[miniMissIndice]);
return {start,miniMiss};
}
if(miniMiss<=errors){
path.push_back(overlap.unitigNumbers[miniMissIndice]);
if(next>start){
return {next,miniMiss};
}
auto res(mapOnRightCache(read , path, {nextOverlapNum,overlap.pos+((uint)nextUnitig.size()-k+1)},listOverlap,ended,start, errors-miniMiss));
return {res.first,res.second+miniMiss};
}
return {start,errors+1};
}
uint Aligner::checkPairCache(const overlapStruct& overlap1, const overlapStruct& overlap2, const string& read, uNumber& number, uint errorsAllowed){
if(overlap2.pos-overlap1.pos<k){
//TODO maybe it is a bad idea ...
return 0;
}
string unitig,subRead(read.substr(overlap1.pos+k-1,overlap2.pos-overlap1.pos-(k-1)));
auto rangeUnitigs1(overlap1.unitig);
auto rangeUnitigs2(overlap2.unitig);
uint minMissMatch(errorsAllowed+1),indice(0);
string unitig1, unitig2;
for(uint i(0); i<rangeUnitigs1.size(); ++i){
unitig1=rangeUnitigs1[i];
for(uint j(0); j<rangeUnitigs2.size(); ++j){
unitig2=rangeUnitigs2[j];
if(unitig2==unitig1){
if(unitig1.size()-2*(k-1)==subRead.size()){
uint missmatch(missmatchNumber(unitig1.substr(k-1,subRead.size()), subRead, errorsAllowed));
if(missmatch<minMissMatch){
minMissMatch=missmatch;
indice=i;
}
}
}
}
}
if(minMissMatch<=errorsAllowed){number=(overlap1.unitigNumbers[indice]);}
return minMissMatch;
}
uint Aligner::mapOnLeftEndExhaustive(const string &read, vector<uNumber>& path, const pair<kmer, uint>& overlap , uint errors){
string unitig, readLeft(read.substr(0,overlap.second));
vector<uNumber> path2keep;
if(readLeft.size()==0){return 0;}
auto rangeUnitigs(getEnd(overlap.first));
uint miniMiss(errors+1),miniMissIndice(9);
int offset(-2);
bool ended(false);
for(uint i(0); i<rangeUnitigs.size(); ++i){
unitig=(rangeUnitigs[i].first);
//case the rest of the read is too small
if(readLeft.size()+k-1 <= unitig.size()){
uint miss(missmatchNumber(unitig.substr(unitig.size()-readLeft.size()-k+1,readLeft.size()), readLeft, errors));
if(miss<miniMiss){
miniMiss=miss;
miniMissIndice=i;
offset=unitig.size()-readLeft.size()-k+1;
ended=true;
}
}else{
//case the read is big enough we want to recover a true overlap
uint miss(missmatchNumber(unitig.substr(0,unitig.size()-k+1), readLeft.substr(readLeft.size()-(unitig.size()-k+1)), errors));
if(miss<miniMiss){
kmer overlapNum(str2num(unitig.substr(0,k-1)));
vector<uNumber> possiblePath;
miss+=mapOnLeftEndExhaustive(read , possiblePath, {overlapNum,overlap.second-(unitig.size()-k+1)}, errors-miss);
if(miss<miniMiss){
path2keep=possiblePath;
miniMiss=miss;
miniMissIndice=i;
offset=-1;
ended=false;
}
}
}
}
if (miniMiss<=errors){
if(ended){
path.push_back(offset);
}else{
path.insert(path.end(), path2keep.begin(),path2keep.end());
}
path.push_back(rangeUnitigs[miniMissIndice].second);
}
return miniMiss;
}
uint Aligner::mapOnLeftEndGreedy(const string &read, vector<uNumber>& path, const pair<kmer, uint>& overlap , uint errors){
// if(overlap.second==0){path.push_back(0);return 0;}
string unitig,readLeft(read.substr(0,overlap.second)),nextUnitig;
auto rangeUnitigs(getEnd(overlap.first));
uint miniMiss(errors+1),miniMissIndice(9);
bool ended(false);
int offset(0);
kmer nextOverlap(0);
for(uint i(0); i<rangeUnitigs.size(); ++i){
unitig=(rangeUnitigs[i].first);
//case the rest of the read is too small
if(readLeft.size()+k-1 <= unitig.size()){
uint miss(missmatchNumber(unitig.substr(unitig.size()-readLeft.size()-k+1,readLeft.size()), readLeft, errors));
if(miss<miniMiss){
miniMiss=miss;
miniMissIndice=i;
ended=true;
offset=unitig.size()-readLeft.size()-k+1;
}
}else{
//case the read is big enough we want to recover a true overlap
uint miss(missmatchNumber(unitig.substr(0,unitig.size()-k+1), readLeft.substr(readLeft.size()-(unitig.size()-k+1)), errors));
if(miss<miniMiss){
kmer overlapNum(str2num(unitig.substr(0,k-1)));
if(miss<miniMiss){
ended=false;
miniMiss=miss;
miniMissIndice=i;
nextUnitig=unitig;
nextOverlap=overlapNum;
}
}
}
}
if (miniMiss<=errors){
if(ended){
path.push_back(offset);
path.push_back(rangeUnitigs[miniMissIndice].second);
return miniMiss;
}
miniMiss+=mapOnLeftEndGreedy(read , path, {nextOverlap,overlap.second-(nextUnitig.size()-k+1)}, errors-miniMiss);
path.push_back(rangeUnitigs[miniMissIndice].second);
}
return miniMiss;
}
uint Aligner::checkEndGreedy(const string& read, pair<kmer, uint>& overlap, vector<uNumber>& path, uint errors){
string readLeft(read.substr(overlap.second+k-1)),unitig,nextUnitig;
if(readLeft.size()<=k-1){return 0;}
auto rangeUnitigs(getBegin(overlap.first));
uint minMiss(errors+1),indiceMinMiss(9);
bool ended(false);
kmer nextOverlap(0);
for(uint i(0); i<rangeUnitigs.size(); ++i){
unitig=(rangeUnitigs[i].first);
if(unitig.size()-k+1>=readLeft.size()){
uint miss(missmatchNumber(unitig.substr(k-1,readLeft.size()),readLeft, errors));
if(miss<minMiss){
minMiss=miss;
indiceMinMiss=i;
ended=true;
}
}else{
uint miss(missmatchNumber(unitig.substr(k-1),readLeft.substr(0,unitig.size()-k+1), errors));
if(miss<minMiss){
if(miss<minMiss){
kmer overlapNum(str2num(unitig.substr(unitig.size()-k+1,k-1)));
minMiss=miss;
indiceMinMiss=i;
nextOverlap=overlapNum;
nextUnitig=unitig;
}
}
}
}
if(minMiss<=errors){
path.push_back(rangeUnitigs[indiceMinMiss].second);
if(ended){
return minMiss;
}
minMiss+=mapOnRightEndGreedy(read, path, {nextOverlap,overlap.second+(nextUnitig.size()-k+1)},errors-minMiss);
if(minMiss<=errors){
successMR++;
}
}
return minMiss;
}
uint Aligner::checkPair(const pair<kmer, uint>& overlap1, const pair<kmer, uint>& overlap2, const string& read, uNumber& number, uint errorsAllowed){
if(overlap2.second-overlap1.second<k){
int32_t positionget1, positionget2;
auto rangeUnitigs1(getBegin(overlap1.first));
auto rangeUnitigs2(getEnd(overlap2.first));
for(uint i(0); i<rangeUnitigs1.size(); ++i){
positionget1=rangeUnitigs1[i].second;
for(uint j(0); j<rangeUnitigs2.size(); ++j){
positionget2=rangeUnitigs2[j].second;
if(positionget2==positionget1){
number=positionget1;
return 0;
}
}
}
return errorsAllowed+1;
}
string unitig,subRead(read.substr(overlap1.second+k-1,overlap2.second-overlap1.second-(k-1)));
auto rangeUnitigs1(getBegin(overlap1.first));
auto rangeUnitigs2(getEnd(overlap2.first));
uint minMissMatch(errorsAllowed+1),indice(0);
int32_t positionget1, positionget2;
for(uint i(0); i<rangeUnitigs1.size(); ++i){
positionget1=rangeUnitigs1[i].second;
for(uint j(0); j<rangeUnitigs2.size(); ++j){
positionget2=rangeUnitigs2[j].second;
if(positionget2==positionget1){
unitig=getUnitig(rangeUnitigs1[i].second);
if(unitig.size()-2*(k-1)==subRead.size()){
uint missmatch(missmatchNumber(unitig.substr(k-1,subRead.size()), subRead, errorsAllowed));
if(missmatch<minMissMatch){
minMissMatch=missmatch;
indice=i;
}
}
}
}
}
if(minMissMatch<=errorsAllowed){number=(rangeUnitigs1[indice].second);}
return minMissMatch;
}