int ReadAlign::oneRead() {//process one read: load, map, write
//load read name, sequence, quality from the streams into internal arrays
int readStatus[2];
readStatus[0]=readLoad(*(readInStream[0]), P, 0, readLength[0], readLengthOriginal[0], readNameMates[0], Read0[0], Read1[0], Qual0[0], Qual1[0], clip3pNtotal[0], clip5pNtotal[0], clip3pAdapterN[0], iReadAll, readFilesIndex, readFilter, readNameExtra[0]);
if (P.readNmates==2) {//read the 2nd mate
readStatus[1]=readLoad(*(readInStream[1]), P, 1, readLength[1], readLengthOriginal[1], readNameMates[1], Read0[1], Read1[0]+readLength[0]+1, Qual0[1], Qual1[0]+readLength[0]+1, clip3pNtotal[1], clip5pNtotal[1], clip3pAdapterN[1], iReadAll, readFilesIndex, readFilter, readNameExtra[1]);
if (readStatus[0]!=readStatus[1]) {
ostringstream errOut;
errOut << "EXITING because of FATAL ERROR: Read1 and Read2 are not consistent, reached the end of the one before the other one\n";
errOut << "SOLUTION: Check you your input files: they may be corrupted\n";
exitWithError(errOut.str(),std::cerr, P.inOut->logMain, EXIT_CODE_INPUT_FILES, P);
} else if (readStatus[0]==-1) {//finished with the stream
return -1;
};
//combine two reads together
Lread=readLength[0]+readLength[1]+1;
readLengthPairOriginal=readLengthOriginal[0]+readLengthOriginal[1]+1;
if (Lread>DEF_readSeqLengthMax) {
ostringstream errOut;
errOut << "EXITING because of FATAL ERROR in reads input: Lread of the pair = " << Lread << " while DEF_readSeqLengthMax=" << DEF_readSeqLengthMax <<endl;
errOut << "Read Name="<<readNameMates[0]<<endl;
errOut << "SOLUTION: increase DEF_readSeqLengthMax in IncludeDefine.h and re-compile STAR"<<endl<<flush;
exitWithError(errOut.str(),std::cerr, P.inOut->logMain, EXIT_CODE_INPUT_FILES, P);
};
Read1[0][readLength[0]]=MARK_FRAG_SPACER_BASE; //marker for spacer base
Qual1[0][readLength[0]]=0;
complementSeqNumbers(Read1[0]+readLength[0]+1,Read1[0]+readLength[0]+1,readLength[1]); //returns complement of Reads[ii]
for (uint ii=0;ii<readLength[1]/2;ii++) {
swap(Read1[0][Lread-ii-1],Read1[0][ii+readLength[0]+1]); //reverse complement
swap(Qual1[0][Lread-ii-1],Qual1[0][ii+readLength[0]+1]); //reverse complement ??? was Qualof the second mate populated
};
} else {//1 mate
if (readStatus[0]==-1) {//finished with the stream
return -1;
};
Lread=readLength[0];
readLengthPairOriginal=readLengthOriginal[0];
readLength[1]=0;
};
readFileType=readStatus[0];
complementSeqNumbers(Read1[0],Read1[1],Lread); //returns complement of Reads[ii]
for (uint ii=0;ii<Lread;ii++) {//reverse
Read1[2][Lread-ii-1]=Read1[1][ii];
Qual1[1][Lread-ii-1]=Qual1[0][ii];
};
statsRA.readN++;
statsRA.readBases += readLength[0]+readLength[1];
//max number of mismatches allowed for this read
outFilterMismatchNmaxTotal=min(P.outFilterMismatchNmax, (uint) (P.outFilterMismatchNoverReadLmax*(readLength[0]+readLength[1])));
//map the read
mapOneRead();
peOverlapMergeMap();
multMapSelect();
mappedFilter();
if (!peOv.yes) {//if the alignment was not mates merged - otherwise the chimeric detection was already done
chimericDetection();
};
if (P.pCh.out.bam && chimRecord) {//chimeric alignment was recorded in main BAM files, and it contains the representative portion, so non-chimeric aligmnent is not output
return 0;
};
waspMap();
#ifdef OFF_BEFORE_OUTPUT
#warning OFF_BEFORE_OUTPUT
return 0;
#endif
//write out alignments
outputAlignments();
return 0;
};
void sjdbBuildIndex (Parameters *P, Parameters *P1, char *Gsj, char *G, PackedArray &SA, PackedArray &SA2, PackedArray &SAi) {
#define SPACER_CHAR GENOME_spacingChar
if (P->sjdbN==0)
{//no junctions to insert
return;
};
time_t rawtime;
time ( &rawtime );
P->inOut->logMain << timeMonthDayTime(rawtime) << " ..... Inserting junctions into the genome indices" <<endl;
*P->inOut->logStdOut << timeMonthDayTime(rawtime) << " ..... Inserting junctions into the genome indices" <<endl;
uint nGsj=P->sjdbLength*P->sjdbN;
for (uint ii=1; ii<=P->sjdbN; ii++)
{
Gsj[ii*P->sjdbLength-1]=SPACER_CHAR; //to make sure this is > than any genome char
};
Gsj[nGsj*2]=SPACER_CHAR+1;//mark the end of the text
for (uint ii=0; ii<nGsj; ii++) {//reverse complement junction sequences
Gsj[nGsj*2-1-ii]=Gsj[ii]<4 ? 3-Gsj[ii] : Gsj[ii]; //reverse complement
};
char* G1c=new char[nGsj*2+1];
complementSeqNumbers(Gsj, G1c, nGsj*2+1);
uint32* oldSJind=new uint32[P1->sjdbN];
// uint nIndicesSJ1=P->sjdbOverhang;
uint nIndicesSJ1=P->sjdbLength;//keep all indices - this is pre-2.4.1 of generating the genome
uint64* indArray=new uint64[2*P->sjdbN*(nIndicesSJ1+1)*2];//8+4 bytes for SA index and index in the genome * nJunction * nIndices per junction * 2 for reverse compl
uint64 sjNew=0;
#pragma omp parallel num_threads(P->runThreadN)
#pragma omp for schedule (dynamic,1000) reduction(+:sjNew)
for (uint isj=0; isj<2*P->sjdbN; isj++) {//find insertion points for each of the sequences
char** seq1=new char*[2];
seq1[0]=Gsj+isj*P->sjdbLength;
seq1[1]=G1c+isj*P->sjdbLength;
uint isj1=isj<P->sjdbN ? isj : 2*P->sjdbN-1-isj;
int sjdbInd = P1->sjdbN==0 ? -1 : binarySearch2(P->sjdbStart[isj1],P->sjdbEnd[isj1],P1->sjdbStart,P1->sjdbEnd,P1->sjdbN);
if (sjdbInd<0)
{//count new junctions
++sjNew;
} else
{//record new index of the old junctions
oldSJind[sjdbInd]=isj1;
};
for (uint istart1=0; istart1<nIndicesSJ1;istart1++) {
uint istart=istart1;
// uint istart=isj<P->sjdbN ? istart1 : istart1+1; //for rev-compl junction, shift by one base to start with the 1st non-spacer base
uint ind1=2*(isj*nIndicesSJ1+istart1);
if (sjdbInd>=0 || seq1[0][istart]>3)
{//no index for already included junctions, or suffices starting with N
indArray[ind1]=-1;
} else
{
//indArray[ind1] = suffixArraySearch(seq1, istart, P->sjdbLength-istart1, G, SA, true, 0, P->nSA-1, 0, P) ;
indArray[ind1] = suffixArraySearch(seq1, istart, 10000, G, SA, true, 0, P->nSA-1, 0, P) ;
indArray[ind1+1] = isj*P->sjdbLength+istart;
};
};
};
// for (int ii=0;ii<P1->sjdbN;ii++) {if ( oldSJind[ii]==0){cout <<ii<<endl;};};
sjNew = sjNew/2;//novel junctions were double counted on two strands
time ( &rawtime );
P->inOut->logMain << timeMonthDayTime(rawtime) << " Finished SA search: number of new junctions=" << sjNew <<", old junctions="<<P->sjdbN-sjNew<<endl;
uint nInd=0;//true number of new indices
for (uint ii=0; ii<2*P->sjdbN*nIndicesSJ1; ii++) {//remove entries that cannot be inserted, this cannot be done in the parallel cycle above
if (indArray[ii*2]!= (uint) -1) {
indArray[nInd*2]=indArray[ii*2];
indArray[nInd*2+1]=indArray[ii*2+1];
++nInd;
};
};
globalGsj=Gsj;
qsort((void*) indArray, nInd, 2*sizeof(uint64), funCompareUintAndSuffixes);
time ( &rawtime );
P->inOut->logMain << timeMonthDayTime(rawtime) << " Finished sorting SA indicesL nInd="<<nInd <<endl;
indArray[2*nInd]=-999; //mark the last junction
indArray[2*nInd+1]=-999; //mark the last junction
P->nGenome=P->chrStart[P->nChrReal]+nGsj;
P->nSA+=nInd;
uint GstrandBit1 = (uint) floor(log(P->nGenome)/log(2))+1;
if (GstrandBit1<32) GstrandBit1=32; //TODO: use simple access function for SA
if ( GstrandBit1 != P->GstrandBit)
{//too many junctions were added - GstrandBit changed
ostringstream errOut;
errOut << "EXITING because of FATAL ERROR: cannot insert junctions on the fly because of strand GstrandBit problem\n";
errOut << "SOLUTION: please contact STAR author at https://groups.google.com/forum/#!forum/rna-star\n";
exitWithError(errOut.str(),std::cerr, P->inOut->logMain, EXIT_CODE_GENOME_FILES, *P);
};
SA2.defineBits(P->GstrandBit+1,P->nSA);
uint nGsjNew=sjNew*P->sjdbLength; //this is the actual number of bytes added to the genome, while nGsj is the total size of all junctions
uint N2bit= 1LLU << P->GstrandBit;
uint strandMask=~N2bit;
//testing
// PackedArray SAo;
// SAo.defineBits(P->GstrandBit+1,P->nSA);
// SAo.allocateArray();
// ifstream oldSAin("./DirTrue/SA");
// oldSAin.read(SAo.charArray,SAo.lengthByte);
// oldSAin.close();
uint isj=0, isa2=0;
for (uint isa=0;isa<P1->nSA;isa++) {
//testing
// if (isa2>0 && SA2[isa2-1]!=SAo[isa2-1]) {
// cout <<isa2 <<" "<< SA2[isa2-1]<<" "<<SAo[isa2-1]<<endl;
// };
// if (isa==69789089)
// {
// cout <<isa;
// };
uint ind1=SA[isa];
if ( (ind1 & N2bit)>0 )
{//- strand
uint ind1s = P1->nGenome - (ind1 & strandMask);
if (ind1s>P->chrStart[P->nChrReal])
{//this index was an old sj, may need to shift it
uint sj1 = (ind1s-P->chrStart[P->nChrReal])/P->sjdbLength;//old junction index
ind1s += (oldSJind[sj1]-sj1)*P->sjdbLength;
ind1 = (P->nGenome - ind1s) | N2bit;
} else
{
ind1+=nGsjNew; //reverse complementary indices are all shifted by the length of junctions
};
} else
{//+ strand
if (ind1>P->chrStart[P->nChrReal])
{//this index was an old sj, may need to shift it
uint sj1 = (ind1-P->chrStart[P->nChrReal])/P->sjdbLength;//old junction index
ind1 += (oldSJind[sj1]-sj1)*P->sjdbLength;
};
};
SA2.writePacked(isa2,ind1); //TODO make sure that the first sj index is not before the first array index
++isa2;
while (isa==indArray[isj*2]) {//insert sj index after the existing index
uint ind1=indArray[isj*2+1];
if (ind1<nGsj) {
ind1+=P->chrStart[P->nChrReal];
} else {//reverse strand
ind1=(ind1-nGsj) | N2bit;
};
SA2.writePacked(isa2,ind1);
++isa2; ++isj;
};
};
time ( &rawtime );
P->inOut->logMain << timeMonthDayTime(rawtime) << " Finished inserting junction indices" <<endl;
//SAi insertions
for (uint iL=0; iL < P->genomeSAindexNbases; iL++) {
uint iSJ=0;
uint ind0=P->genomeSAindexStart[iL]-1;//last index that was present in the old genome
for (uint ii=P->genomeSAindexStart[iL];ii<P->genomeSAindexStart[iL+1]; ii++) {//scan through the longest index
uint iSA1=SAi[ii];
uint iSA2=iSA1 & P->SAiMarkNmask & P->SAiMarkAbsentMask;
if ( iSJ<nInd && (iSA1 & P->SAiMarkAbsentMaskC)>0 )
{//index missing from the old genome
uint iSJ1=iSJ;
int64 ind1=funCalcSAi(Gsj+indArray[2*iSJ+1],iL);
while (ind1 < (int64)(ii-P->genomeSAindexStart[iL]) && indArray[2*iSJ]<iSA2) {
++iSJ;
ind1=funCalcSAi(Gsj+indArray[2*iSJ+1],iL);
};
if (ind1 == (int64)(ii-P->genomeSAindexStart[iL]) ) {
SAi.writePacked(ii,indArray[2*iSJ]+iSJ+1);
for (uint ii0=ind0+1; ii0<ii; ii0++) {//fill all the absent indices with this value
SAi.writePacked(ii0,(indArray[2*iSJ]+iSJ+1) | P->SAiMarkAbsentMaskC);
};
++iSJ;
ind0=ii;
} else {
iSJ=iSJ1;
};
} else
{//index was present in the old genome
while (iSJ<nInd && indArray[2*iSJ]+1<iSA2) {//for this index insert "smaller" junctions
++iSJ;
};
while (iSJ<nInd && indArray[2*iSJ]+1==iSA2) {//special case, the index falls right behind SAi
if (funCalcSAi(Gsj+indArray[2*iSJ+1],iL) >= (int64) (ii-P->genomeSAindexStart[iL]) ) {//this belongs to the next index
break;
};
++iSJ;
};
SAi.writePacked(ii,iSA1+iSJ);
for (uint ii0=ind0+1; ii0<ii; ii0++) {//fill all the absent indices with this value
SAi.writePacked(ii0,(iSA2+iSJ) | P->SAiMarkAbsentMaskC);
};
ind0=ii;
};
};
};
// time ( &rawtime ); cout << timeMonthDayTime(rawtime) << "SAi first" <<endl;
for (uint isj=0;isj<nInd;isj++) {
int64 ind1=0;
for (uint iL=0; iL < P->genomeSAindexNbases; iL++) {
uint g=(uint) Gsj[indArray[2*isj+1]+iL];
ind1 <<= 2;
if (g>3) {//this iSA contains N, need to mark the previous
for (uint iL1=iL; iL1 < P->genomeSAindexNbases; iL1++) {
ind1+=3;
int64 ind2=P->genomeSAindexStart[iL1]+ind1;
for (; ind2>=0; ind2--) {//find previous index that is not absent
if ( (SAi[ind2] & P->SAiMarkAbsentMaskC)==0 ) {
break;
};
};
SAi.writePacked(ind2,SAi[ind2] | P->SAiMarkNmaskC);
ind1 <<= 2;
};
break;
} else {
ind1 += g;
};
};
};
time ( &rawtime );
P->inOut->logMain << timeMonthDayTime(rawtime) << " Finished SAi" <<endl;
//change parameters, most parameters are already re-defined in sjdbPrepare.cpp
SA.defineBits(P->GstrandBit+1,P->nSA);//same as SA2
SA.pointArray(SA2.charArray);
P->nSAbyte=SA.lengthByte;
P->sjGstart=P->chrStart[P->nChrReal];
memcpy(G+P->chrStart[P->nChrReal],Gsj, nGsj);
/* testing
PackedArray SAio=SAi;
SAio.allocateArray();
ifstream oldSAiin("./DirTrue/SAindex");
// oldSAin.read(SAio.charArray,8*(P->genomeSAindexNbases+2));//skip first bytes
oldSAiin.read(SAio.charArray,SAio.lengthByte);
oldSAiin.close();
// for (uint ii=0;ii<P->nSA;ii++) {
// if (SA2[ii]!=SAo[ii]) {
// cout <<ii <<" "<< SA2[ii]<<" "<<SAo[ii]<<endl;
// };
// };
for (uint iL=0; iL < P->genomeSAindexNbases; iL++) {
for (uint ii=P->genomeSAindexStart[iL];ii<P->genomeSAindexStart[iL+1]; ii++) {//scan through the longets index
if ( SAio[ii]!=SAi[ii] ) {
cout <<ii<<" "<<SAio[ii]<<" "<<SAi[ii]<<endl;
};
};
};
*/
/*
ofstream genomeOut("/home/dobin/Genome");
fstreamWriteBig(genomeOut,G,P->nGenome+nGsj,"777","777",P);
genomeOut.close();
genomeOut.open("/home/dobin/SA");
fstreamWriteBig(genomeOut,SA2.charArray,SA2.lengthByte,"777","777",P);
genomeOut.close();
*/
delete [] indArray;
delete [] G1c;
delete [] oldSJind;
};
void ReadAlign::peMergeMates() {
uint s1=localSearchNisMM(Read1[0],readLength[0],Read1[0]+readLength[0]+1,readLength[1],P.peOverlap.MMp);
uint s0=localSearchNisMM(Read1[0]+readLength[0]+1,readLength[1],Read1[0],readLength[0],P.peOverlap.MMp);
uint o1=min(readLength[1],readLength[0]-s1);
uint o0=min(readLength[0],readLength[1]-s0);
peOv.nOv=max(o0,o1);
if (peOv.nOv<P.peOverlap.NbasesMin) {//overlap is smaller than minimum allowed
peOv.nOv=0;
return;
};
if (o1>=o0) {
peOv.mateStart[0]=0;
peOv.mateStart[1]=s1;
if (o1<readLength[1]) {//otherwise, if o1==readLength[1], read2 is entirely contained in read1
//move unoverlapped portion of read2 to the end of read1
memmove(Read1[0]+readLength[0], Read1[0]+readLength[0]+1+o1, readLength[1]-o1);
};
} else {
peOv.mateStart[1]=0;
peOv.mateStart[0]=s0;
memmove(Read1[0]+Lread, Read1[0], readLength[0]);//temp move 0
memmove(Read1[0], Read1[0]+readLength[0]+1, readLength[1]); //move 1 into 0
if (o0<readLength[0]) {
memmove(Read1[0]+readLength[1], Read1[0]+Lread+o0, readLength[0]-o0); //move 0 into 1
};
};
//uint nMM=0;
//for (uint ii=peOv.ovS; ii<readLength[0]; ii++) {//check for MM in the overlap area
// if (Read1[0][ii]!=Read1[0][ii-peOv.ovS+readLength[0]+1]) {
// Read1[0][ii]=4; //replace mismatched base with N
// ++nMM;
// };
//};
Lread=Lread-peOv.nOv-1;
readLength[0]=Lread;
readLength[1]=0;
readLengthOriginal[0]=Lread;
readLengthOriginal[1]=0;
readNmates=1;
//fill Read1[1,2]
complementSeqNumbers(Read1[0],Read1[1],Lread); //returns complement of Reads[ii]
for (uint ii=0;ii<Lread;ii++) {//reverse
Read1[2][Lread-ii-1]=Read1[1][ii];
if (Read1[1][ii]<4) {
Qual1[0][ii]=1;
Qual1[1][Lread-ii-1]=1;
} else {
Qual1[0][ii]=0;
Qual1[1][Lread-ii-1]=0;
};
};
return;
};
uint insertSeqSA(PackedArray & SA, PackedArray & SA1, PackedArray & SAi, char * G, char * G1, uint64 nG, uint64 nG1, uint64 nG2, Parameters * P)
{//insert new sequences into the SA
uint GstrandBit1 = (uint) floor(log(nG+nG1)/log(2))+1;
if (GstrandBit1<32) GstrandBit1=32; //TODO: use simple access function for SA
if ( GstrandBit1+1 != SA.wordLength)
{//sequence is too long - GstrandBit changed
ostringstream errOut;
errOut << "EXITING because of FATAL ERROR: cannot insert sequence on the fly because of strand GstrandBit problem\n";
errOut << "SOLUTION: please contact STAR author at https://groups.google.com/forum/#!forum/rna-star\n";
exitWithError(errOut.str(),std::cerr, P->inOut->logMain, EXIT_CODE_GENOME_FILES, *P);
};
uint N2bit= 1LLU << (SA.wordLength-1);
uint strandMask=~N2bit;
for (uint64 isa=0;isa<SA.length; isa++)
{
uint64 ind1=SA[isa];
if ( (ind1 & N2bit)>0 )
{//- strand
if ( (ind1 & strandMask)>=nG2 )
{//the first nG bases
ind1+=nG1; //reverse complementary indices are all shifted by the length of the sequence
SA.writePacked(isa,ind1);
};
} else
{//+ strand
if ( ind1>=nG )
{//the last nG2 bases
ind1+=nG1; //reverse complementary indices are all shifted by the length of the sequence
SA.writePacked(isa,ind1);
};
};
};
char** seq1=new char*[2];
#define GENOME_endFillL 16
char* seqq=new char [4*nG1+3*GENOME_endFillL];//ends shouldbe filled with 5 to mark boundaries
seq1[0]=seqq+GENOME_endFillL;//TODO: avoid defining an extra array, use reverse search
seq1[1]=seqq+2*GENOME_endFillL+2*nG1;
memset(seqq,GENOME_spacingChar,GENOME_endFillL);
memset(seqq+2*nG1+GENOME_endFillL,GENOME_spacingChar,GENOME_endFillL);
memset(seqq+4*nG1+2*GENOME_endFillL,GENOME_spacingChar,GENOME_endFillL);
memcpy(seq1[0], G1, nG1);
for (uint ii=0; ii<nG1; ii++)
{//reverse complement sequence
seq1[0][2*nG1-1-ii]=seq1[0][ii]<4 ? 3-seq1[0][ii] : seq1[0][ii];
};
complementSeqNumbers(seq1[0], seq1[1], 2*nG1);//complement
uint64* indArray=new uint64[nG1*2*2+2];// for each base, 1st number - insertion place in SA, 2nd number - index, *2 for reverse compl
#pragma omp parallel num_threads(P->runThreadN)
#pragma omp for schedule (dynamic,1000)
for (uint ii=0; ii<2*nG1; ii++) {//find insertion points for each of the sequences
if (seq1[0][ii]>3)
{//no index for suffices starting with N
indArray[ii*2]=-1;
} else
{
indArray[ii*2] = suffixArraySearch1(seq1, ii, 10000, G, nG, SA, (ii<nG1 ? true:false), 0, SA.length-1, 0, P) ;
indArray[ii*2+1] = ii;
};
};
uint64 nInd=0;//true number of new indices
for (uint ii=0; ii<2*nG1; ii++) {//remove entries that cannot be inserted, this cannot be done in the parallel cycle above
if (indArray[ii*2]!= (uint) -1) {
indArray[nInd*2]=indArray[ii*2];
indArray[nInd*2+1]=indArray[ii*2+1];
++nInd;
};
};
time_t rawtime;
time ( &rawtime );
P->inOut->logMain << timeMonthDayTime(rawtime) << " Finished SA search, number of new SA indices = "<<nInd<<endl;
globalGenomeArray=seq1[0];
qsort((void*) indArray, nInd, 2*sizeof(uint64), funCompareUintAndSuffixes);
time ( &rawtime );
P->inOut->logMain << timeMonthDayTime(rawtime) << " Finished sorting SA indices"<<endl;
indArray[2*nInd]=-999; //mark the last junction
indArray[2*nInd+1]=-999; //mark the last junction
SA1.defineBits(SA.wordLength,SA.length+nInd);
/*testing
PackedArray SAo;
SAo.defineBits(P->GstrandBit+1,P->nSA+nInd);
SAo.allocateArray();
ifstream oldSAin("./DirTrue/SA");
oldSAin.read(SAo.charArray,SAo.lengthByte);
oldSAin.close();
*/
uint isa1=0, isa2=0;
for (uint isa=0;isa<SA.length;isa++) {
while (isa==indArray[isa1*2]) {//insert new index before the existing index
uint ind1=indArray[isa1*2+1];
if (ind1<nG1) {
ind1+=nG;
} else {//reverse strand
ind1=(ind1-nG1+nG2) | N2bit;
};
SA1.writePacked(isa2,ind1);
/*testing
if (SA1[isa2]!=SAo[isa2]) {
cout <<isa2 <<" "<< SA1[isa2]<<" "<<SAo[isa2]<<endl;
//sleep(100);
};
*/
++isa2; ++isa1;
};
SA1.writePacked(isa2,SA[isa]); //TODO make sure that the first sj index is not before the first array index
/*testing
if (SA1[isa2]!=SAo[isa2]) {
cout <<isa2 <<" "<< SA1[isa2]<<" "<<SAo[isa2]<<endl;
//sleep(100);
};
*/
++isa2;
};
for (;isa1<nInd;isa1++)
{//insert the last indices
uint ind1=indArray[isa1*2+1];
if (ind1<nG1)
{
ind1+=nG;
} else
{//reverse strand
ind1=(ind1-nG1+nG2) | N2bit;
};
SA1.writePacked(isa2,ind1);
++isa2;
};
time ( &rawtime );
P->inOut->logMain << timeMonthDayTime(rawtime) << " Finished inserting SA indices" <<endl;
// //SAi insertions
// for (uint iL=0; iL < P->genomeSAindexNbases; iL++) {
// uint iSeq=0;
// uint ind0=P->genomeSAindexStart[iL]-1;//last index that was present in the old genome
// for (uint ii=P->genomeSAindexStart[iL];ii<P->genomeSAindexStart[iL+1]; ii++) {//scan through the longest index
// if (ii==798466)
// cout <<ii;
//
// uint iSA1=SAi[ii];
// uint iSA2=iSA1 & P->SAiMarkNmask & P->SAiMarkAbsentMask;
//
// if ( iSeq<nInd && (iSA1 & P->SAiMarkAbsentMaskC)>0 )
// {//index missing from the old genome
// uint iSeq1=iSeq;
// int64 ind1=funCalcSAi(seq1[0]+indArray[2*iSeq+1],iL);
// while (ind1 < (int64)(ii-P->genomeSAindexStart[iL]) && indArray[2*iSeq]<iSA2) {
// ++iSeq;
// ind1=funCalcSAi(seq1[0]+indArray[2*iSeq+1],iL);
// };
// if (ind1 == (int64)(ii-P->genomeSAindexStart[iL]) ) {
// SAi.writePacked(ii,indArray[2*iSeq]+iSeq+1);
// for (uint ii0=ind0+1; ii0<ii; ii0++) {//fill all the absent indices with this value
// SAi.writePacked(ii0,(indArray[2*iSeq]+iSeq+1) | P->SAiMarkAbsentMaskC);
// };
// ++iSeq;
// ind0=ii;
// } else {
// iSeq=iSeq1;
// };
// } else
// {//index was present in the old genome
// while (iSeq<nInd && indArray[2*iSeq]+1<iSA2) {//for this index insert "smaller" junctions
// ++iSeq;
// };
//
// while (iSeq<nInd && indArray[2*iSeq]+1==iSA2) {//special case, the index falls right behind SAi
// if (funCalcSAi(seq1[0]+indArray[2*iSeq+1],iL) >= (int64) (ii-P->genomeSAindexStart[iL]) ) {//this belongs to the next index
// break;
// };
// ++iSeq;
// };
//
// SAi.writePacked(ii,iSA1+iSeq);
//
// for (uint ii0=ind0+1; ii0<ii; ii0++) {//fill all the absent indices with this value
// SAi.writePacked(ii0,(iSA2+iSeq) | P->SAiMarkAbsentMaskC);
// };
// ind0=ii;
// };
// };
//
// };
// // time ( &rawtime ); cout << timeMonthDayTime(rawtime) << "SAi first" <<endl;
//
// for (uint isj=0;isj<nInd;isj++) {
// int64 ind1=0;
// for (uint iL=0; iL < P->genomeSAindexNbases; iL++) {
// uint g=(uint) seq1[0][indArray[2*isj+1]+iL];
// ind1 <<= 2;
// if (g>3) {//this iSA contains N, need to mark the previous
// for (uint iL1=iL; iL1 < P->genomeSAindexNbases; iL1++) {
// ind1+=3;
// int64 ind2=P->genomeSAindexStart[iL1]+ind1;
// for (; ind2>=0; ind2--) {//find previous index that is not absent
// if ( (SAi[ind2] & P->SAiMarkAbsentMaskC)==0 ) {
// break;
// };
// };
// SAi.writePacked(ind2,SAi[ind2] | P->SAiMarkNmaskC);
// ind1 <<= 2;
// };
// break;
// } else {
// ind1 += g;
// };
// };
// };
// time ( &rawtime );
// P->inOut->logMain << timeMonthDayTime(rawtime) << " Finished SAi" <<endl;
//
// /* testing
// PackedArray SAio=SAi;
// SAio.allocateArray();
// ifstream oldSAiin("./DirTrue/SAindex");
// oldSAiin.read(SAio.charArray,8*(P->genomeSAindexNbases+2));//skip first bytes
// oldSAiin.read(SAio.charArray,SAio.lengthByte);
// oldSAiin.close();
//
// for (uint iL=0; iL < P->genomeSAindexNbases; iL++) {
// for (uint ii=P->genomeSAindexStart[iL];ii<P->genomeSAindexStart[iL+1]; ii++) {//scan through the longets index
// if ( SAio[ii]!=SAi[ii] ) {
// cout <<iL<<" "<<ii<<" "<<SAio[ii]<<" "<<SAi[ii]<<endl;
// };
// };
// };
// */
//change parameters, most parameters are already re-defined in sjdbPrepare.cpp
SA.defineBits(P->GstrandBit+1,SA.length+nInd);//same as SA2
SA.pointArray(SA1.charArray);
P->nSA=SA.length;
P->nSAbyte=SA.lengthByte;
//generate SAi
genomeSAindex(G,SA,P,SAi);
time ( &rawtime );
P->inOut->logMain << timeMonthDayTime(rawtime) << " Finished SAi" <<endl;
// P->sjGstart=P->chrStart[P->nChrReal];
// memcpy(G+P->chrStart[P->nChrReal],seq1[0], nseq1[0]);
return nInd;
};
