void calcExpectedEffectiveLengths(ModelType& model) {
int lb, ub, span;
double *pdf = NULL, *cdf = NULL, *clen = NULL; // clen[i] = sigma_{j=1}^{i}pdf[i]*(lb+i)
model.getGLD().copyTo(pdf, cdf, lb, ub, span);
clen = new double[span + 1];
clen[0] = 0.0;
for (int i = 1; i <= span; i++) {
clen[i] = clen[i - 1] + pdf[i] * (lb + i);
}
eel.clear();
eel.resize(M + 1, 0.0);
for (int i = 1; i <= M; i++) {
int totLen = refs.getRef(i).getTotLen();
int fullLen = refs.getRef(i).getFullLen();
int pos1 = max(min(totLen - fullLen + 1, ub) - lb, 0);
int pos2 = max(min(totLen, ub) - lb, 0);
if (pos2 == 0) { eel[i] = 0.0; continue; }
eel[i] = fullLen * cdf[pos1] + ((cdf[pos2] - cdf[pos1]) * (totLen + 1) - (clen[pos2] - clen[pos1]));
assert(eel[i] >= 0);
if (eel[i] < MINEEL) { eel[i] = 0.0; }
}
delete[] pdf;
delete[] cdf;
delete[] clen;
}
inline void PROBerReadModel::update(InMemAlignG* ag_in_mem, InMemAlign* aligns, AlignmentGroup& ag, double noise_frac) {
SEQstring seq;
QUALstring qual;
CIGARstring cigar;
const RefSeq* refseq = NULL;
assert(ag.getSEQ(seq));
if (model_type & 1) assert(ag.getQUAL(qual));
// update noise prob
npro->update(seq, noise_frac);
// update alignment probs
for (int i = 0; i < ag_in_mem->size; ++i) if (aligns[i].frac > 0.0) {
refseq = refs->getRef(aligns[i].tid);
assert(ag.getAlignment(i)->getCIGAR(cigar));
seqmodel->update(aligns[i].frac, '+', aligns[i].pos, refseq, &cigar, &seq, ((model_type & 1) ? &qual : NULL));
}
if (model_type >= 2) {
// paired-end reads
assert(ag.getSEQ(seq, 2));
if (model_type & 1) assert(ag.getQUAL(qual, 2));
// update noise prob
npro->update(seq, noise_frac);
// update alignment probs
for (int i = 0; i < ag_in_mem->size; ++i) if (aligns[i].frac > 0.0) {
refseq = refs->getRef(aligns[i].tid);
assert(ag.getAlignment(i)->getCIGAR(cigar, 2));
assert(aligns[i].fragment_length > 0);
seqmodel->update(aligns[i].frac, '-', refseq->getLen() - aligns[i].pos - aligns[i].fragment_length, refseq, &cigar, &seq, ((model_type & 1) ? &qual : NULL));
}
}
}
inline void PROBerReadModel::setConProbs(InMemAlignG* ag_in_mem, InMemAlign* aligns, AlignmentGroup& ag) {
int seqlen;
SEQstring seq; // seq, qual and cigar must be in each function since we have multiple threads!
QUALstring qual;
CIGARstring cigar;
const RefSeq* refseq = NULL;
// Get read sequences and quality scores
assert(ag.getSEQ(seq));
seqlen = read_length < 0 ? ag.getSeqLength() : read_length;
if (model_type & 1) assert(ag.getQUAL(qual));
// set noise probability
ag_in_mem->noise_conprb = mld1->getProb(seqlen) * npro->getProb(seq);
// set alignment probabilities
for (int i = 0; i < ag_in_mem->size; ++i) if (aligns[i].conprb != -1.0) {
refseq = refs->getRef(aligns[i].tid);
assert(ag.getAlignment(i)->getCIGAR(cigar));
aligns[i].conprb = (aligns[i].fragment_length > 0 ? mld1->getProb(seqlen, aligns[i].fragment_length) : mld1->getProb(seqlen)) * \
seqmodel->getProb('+', aligns[i].pos, refseq, &cigar, &seq, ((model_type & 1) ? &qual : NULL));
}
if (model_type >= 2) {
// paired-end reads
assert(ag.getSEQ(seq, 2));
seqlen = read_length < 0 ? ag.getSeqLength(2) : read_length;
if (model_type & 1) assert(ag.getQUAL(qual, 2));
ag_in_mem->noise_conprb *= mld2->getProb(seqlen) * npro->getProb(seq);
for (int i = 0; i < ag_in_mem->size; ++i) if (aligns[i].conprb != -1.0) {
refseq = refs->getRef(aligns[i].tid);
assert(ag.getAlignment(i)->getCIGAR(cigar, 2));
assert(aligns[i].fragment_length > 0);
aligns[i].conprb *= mld2->getProb(seqlen, aligns[i].fragment_length) * \
seqmodel->getProb('-', refseq->getLen() - aligns[i].pos - aligns[i].fragment_length, refseq, &cigar, &seq, ((model_type & 1) ? &qual : NULL));
}
}
}
inline void PROBerReadModel::simulate(READ_INT_TYPE rid, int tid, int pos, int fragment_length, std::ofstream* out1, std::ofstream* out2) {
int m2pos;
int mateL1, mateL2;
std::string qual1, qual2, cigar1, cigar2, readseq1, readseq2;
// Simulate reads
if (tid == 0) {
cigar1 = cigar2 = "*";
mateL1 = mld1->simulate(sampler);
if (model_type & 1) qd->simulate(sampler, mateL1, qual1);
npro->simulate(sampler, mateL1, readseq1);
if (model_type >= 2) {
mateL2 = mld2->simulate(sampler);
if (model_type & 1) qd->simulate(sampler, mateL2, qual2);
npro->simulate(sampler, mateL2, readseq2);
}
}
else {
const RefSeq* ref = refs->getRef(tid);
mateL1 = mld1->simulate(sampler, fragment_length);
if (model_type & 1) qd->simulate(sampler, mateL1, qual1);
seqmodel->simulate(sampler, mateL1, '+', pos, ref, qual1, cigar1, readseq1);
if (model_type >= 2) {
mateL2 = mld2->simulate(sampler, fragment_length);
if (model_type & 1) qd->simulate(sampler, mateL2, qual2);
m2pos = ref->getLen() - pos - fragment_length;
seqmodel->simulate(sampler, mateL2, '-', m2pos, ref, qual2, cigar2, readseq2);
}
}
// Output reads
(*out1)<< ((model_type & 1) ? '@' : '>') << rid<< '_'<< tid<< '_'<< pos<< '_'<< fragment_length<< '_'<< cigar1<< (model_type >= 2 ? "/1" : "") << std::endl;
(*out1)<< readseq1<< std::endl;
if (model_type & 1) (*out1)<< '+'<< std::endl<< qual1<< std::endl;
if (model_type >= 2) {
(*out2)<< ((model_type & 1) ? '@' : '>') << rid<< '_'<< tid<< '_'<< pos<< '_'<< fragment_length<< '_'<< cigar2<< "/2"<< std::endl;
(*out2)<< readseq2<< std::endl;
if (model_type & 1) (*out2)<< '+'<< std::endl<< qual2<< std::endl;
}
}
void parr_update_duplicates(Ob ** obs, size_t obs_N, PDUP_t dupT, Refs& new_dups)
{
REL_t pHUB = REL_PARENT;
REL_t pEXA = REL_CHILD;
cmp_function_t *cmp = nullptr;
// determine \c cmp function (functional argument)
switch (dupT) {
case PDUP_NAME:
cmp = voidp_ob_cmp_name;
break;
case PDUP_CONTENT:
cmp = voidp_ob_cmp_std;
break;
default: PTODO("Cannot handle dupT:%d (yet)\n", dupT); return; break;
}
qsort_mt(obs, obs_N, sizeof(Ob*), cmp, 0,0);
for (size_t i = 0; i < obs_N;) { // for all obs
size_t j = i+1;
// iterate j to beyond first non-duplicate
for (; j < obs_N; j++) { // for all obs following obs[i]
if (cmp(&obs[i], &obs[j]) != 0) { break; }
}
const size_t dupN = j-i; // multiplicity (2 or more for duplicates)
if (dupN >= 2) { // obsj[i ... i+dupN-1] are duplicates
Dup * pdup = gen::dup(dupT);
new_dups.app(pdup);
for (size_t k = 0; k < dupN; k++) { // for all duplicates
obs[i+k]->net_disconnectM(pHUB); // disconnect from all previous
net_connectS(pEXA, obs[i+k], // connect to new
pHUB, pdup,
true);
}
} else { // obs[i] has no duplicates
obs[i]->net_disconnectM(pHUB); // disconnect from all previous
}
i = j;
}
}
void init(ReadReader<ReadType> **&readers, HitContainer<HitType> **&hitvs, double **&ncpvs, ModelType **&mhps) {
READ_INT_TYPE nReads;
HIT_INT_TYPE nHits;
int rt; // read type
READ_INT_TYPE nrLeft, curnr; // nrLeft : number of reads left, curnr: current number of reads
HIT_INT_TYPE nhT; // nhT : hit threshold per thread
char datF[STRLEN];
int s;
char readFs[2][STRLEN];
ReadIndex *indices[2];
ifstream fin;
readers = new ReadReader<ReadType>*[nThreads];
genReadFileNames(imdName, 1, read_type, s, readFs);
for (int i = 0; i < s; i++) {
indices[i] = new ReadIndex(readFs[i]);
}
for (int i = 0; i < nThreads; i++) {
readers[i] = new ReadReader<ReadType>(s, readFs, refs.hasPolyA(), mparams.seedLen); // allow calculation of calc_lq() function
readers[i]->setIndices(indices);
}
hitvs = new HitContainer<HitType>*[nThreads];
for (int i = 0; i < nThreads; i++) {
hitvs[i] = new HitContainer<HitType>();
}
sprintf(datF, "%s.dat", imdName);
fin.open(datF);
general_assert(fin.is_open(), "Cannot open " + cstrtos(datF) + "! It may not exist.");
fin>>nReads>>nHits>>rt;
general_assert(nReads == N1, "Number of alignable reads does not match!");
general_assert(rt == read_type, "Data file (.dat) does not have the right read type!");
//A just so so strategy for paralleling
nhT = nHits / nThreads;
nrLeft = N1;
curnr = 0;
ncpvs = new double*[nThreads];
for (int i = 0; i < nThreads; i++) {
HIT_INT_TYPE ntLeft = nThreads - i - 1; // # of threads left
general_assert(readers[i]->locate(curnr), "Read indices files do not match!");
while (nrLeft > ntLeft && (i == nThreads - 1 || hitvs[i]->getNHits() < nhT)) {
general_assert(hitvs[i]->read(fin), "Cannot read alignments from .dat file!");
--nrLeft;
if (verbose && nrLeft % 1000000 == 0) { cout<< "DAT "<< nrLeft << " reads left"<< endl; }
}
ncpvs[i] = new double[hitvs[i]->getN()];
memset(ncpvs[i], 0, sizeof(double) * hitvs[i]->getN());
curnr += hitvs[i]->getN();
if (verbose) { cout<<"Thread "<< i<< " : N = "<< hitvs[i]->getN()<< ", NHit = "<< hitvs[i]->getNHits()<< endl; }
}
fin.close();
mhps = new ModelType*[nThreads];
for (int i = 0; i < nThreads; i++) {
mhps[i] = new ModelType(mparams, false); // just model helper
}
probv = new double[M + 1];
countvs = new double*[nThreads];
for (int i = 0; i < nThreads; i++) {
countvs[i] = new double[M + 1];
}
if (verbose) { printf("EM_init finished!\n"); }
}
int main(int argc, char* argv[]) {
ifstream fin;
bool quiet = false;
if (argc < 5) {
printf("Usage : rsem-run-em refName read_type sampleName sampleToken [-p #Threads] [-b samInpType samInpF has_fn_list_? [fn_list]] [-q] [--gibbs-out] [--sampling]\n\n");
printf(" refName: reference name\n");
printf(" read_type: 0 single read without quality score; 1 single read with quality score; 2 paired-end read without quality score; 3 paired-end read with quality score.\n");
printf(" sampleName: sample's name, including the path\n");
printf(" sampleToken: sampleName excludes the path\n");
printf(" -p: number of threads which user wants to use. (default: 1)\n");
printf(" -b: produce bam format output file. (default: off)\n");
printf(" -q: set it quiet\n");
printf(" --gibbs-out: generate output file used by Gibbs sampler. (default: off)\n");
printf(" --sampling: sample each read from its posterior distribution when bam file is generated. (default: off)\n");
printf("// model parameters should be in imdName.mparams.\n");
exit(-1);
}
time_t a = time(NULL);
strcpy(refName, argv[1]);
read_type = atoi(argv[2]);
strcpy(outName, argv[3]);
sprintf(imdName, "%s.temp/%s", argv[3], argv[4]);
sprintf(statName, "%s.stat/%s", argv[3], argv[4]);
nThreads = 1;
genBamF = false;
bamSampling = false;
genGibbsOut = false;
pt_fn_list = pt_chr_list = NULL;
for (int i = 5; i < argc; i++) {
if (!strcmp(argv[i], "-p")) { nThreads = atoi(argv[i + 1]); }
if (!strcmp(argv[i], "-b")) {
genBamF = true;
inpSamType = argv[i + 1][0];
strcpy(inpSamF, argv[i + 2]);
if (atoi(argv[i + 3]) == 1) {
strcpy(fn_list, argv[i + 4]);
pt_fn_list = (char*)(&fn_list);
}
}
if (!strcmp(argv[i], "-q")) { quiet = true; }
if (!strcmp(argv[i], "--gibbs-out")) { genGibbsOut = true; }
if (!strcmp(argv[i], "--sampling")) { bamSampling = true; }
}
general_assert(nThreads > 0, "Number of threads should be bigger than 0!");
verbose = !quiet;
//basic info loading
sprintf(refF, "%s.seq", refName);
refs.loadRefs(refF);
M = refs.getM();
sprintf(groupF, "%s.grp", refName);
gi.load(groupF);
m = gi.getm();
sprintf(tiF, "%s.ti", refName);
transcripts.readFrom(tiF);
sprintf(cntF, "%s.cnt", statName);
fin.open(cntF);
general_assert(fin.is_open(), "Cannot open " + cstrtos(cntF) + "! It may not exist.");
fin>>N0>>N1>>N2>>N_tot;
fin.close();
general_assert(N1 > 0, "There are no alignable reads!");
if ((READ_INT_TYPE)nThreads > N1) nThreads = N1;
//set model parameters
mparams.M = M;
mparams.N[0] = N0; mparams.N[1] = N1; mparams.N[2] = N2;
mparams.refs = &refs;
sprintf(mparamsF, "%s.mparams", imdName);
fin.open(mparamsF);
general_assert(fin.is_open(), "Cannot open " + cstrtos(mparamsF) + "It may not exist.");
fin>> mparams.minL>> mparams.maxL>> mparams.probF;
int val; // 0 or 1 , for estRSPD
fin>>val;
mparams.estRSPD = (val != 0);
fin>> mparams.B>> mparams.mate_minL>> mparams.mate_maxL>> mparams.mean>> mparams.sd;
fin>> mparams.seedLen;
fin.close();
//run EM
switch(read_type) {
case 0 : EM<SingleRead, SingleHit, SingleModel>(); break;
case 1 : EM<SingleReadQ, SingleHit, SingleQModel>(); break;
case 2 : EM<PairedEndRead, PairedEndHit, PairedEndModel>(); break;
case 3 : EM<PairedEndReadQ, PairedEndHit, PairedEndQModel>(); break;
default : fprintf(stderr, "Unknown Read Type!\n"); exit(-1);
}
time_t b = time(NULL);
printTimeUsed(a, b, "EM.cpp");
return 0;
}
int main(int argc, char* argv[]) {
if (argc < 2) {
printf("Usage: PROBer-build-reference refName [--gtf gtfF] [--mapping mappingF] [--allele-specific] [--files num_of_files file_1 file_2 ...] [--n2g-index] [-q]\n");
exit(-1);
}
hasGTF = false;
mappingType = 0;
n2g_idx = false;
int argpos = 2;
while (argpos < argc) {
if (!strcmp(argv[argpos], "--gtf")) {
hasGTF = true;
strcpy(gtfF, argv[++argpos]);
}
if (!strcmp(argv[argpos], "--mapping")) {
mappingType = 1;
mappingPos = ++argpos;
}
if (!strcmp(argv[argpos], "--allele-specific")) mappingType = 2;
if (!strcmp(argv[argpos], "--files")) {
num_files = atoi(argv[++argpos]);
file_pos = argpos + 1; // the position in argv for the first file
argpos += num_files;
}
if (!strcmp(argv[argpos], "--n2g-index")) n2g_idx = true;
if (!strcmp(argv[argpos], "-q")) verbose = false;
++argpos;
}
if (mappingType > 0) loadMappingInfo(mappingType, argv[mappingPos]);
ifstream fin;
string line, gseq, tseq; // gseq, genomic sequence; tseq, transcript sequence
string seqname, gene_id, transcript_id;
if (hasGTF) {
transcripts.setType(0);
assert(mappingType < 2);
parse_gtf_file(gtfF);
M = transcripts.getM();
general_assert(M > 0, "The reference contains no transcripts!");
seqs.assign(M + 1, "");
chrvec.clear();
for (int i = 0; i < num_files; ++i, ++file_pos) {
fin.open(argv[file_pos]);
general_assert(fin.is_open(), "Cannot open " + cstrtos(argv[file_pos]) + "! It may not exist.");
getline(fin, line);
while ((fin) && (line[0] == '>')) {
istringstream strin(line.substr(1));
strin>>seqname;
gseq = "";
while((getline(fin, line)) && (line[0] != '>')) {
gseq += line;
}
assert(gseq.length() > 0);
sn2tr_iter = sn2tr.find(seqname);
if (sn2tr_iter == sn2tr.end()) continue;
chrvec.push_back(ChrInfo(seqname, gseq.length()));
vector<int>& vec = sn2tr_iter->second;
int s = vec.size();
for (int j = 0; j < s; ++j) {
assert(vec[j] > 0 && vec[j] <= M);
transcripts.getTranscriptAt(vec[j]).extractSeq(gseq, seqs[vec[j]]);
}
}
fin.close();
if (verbose) { printf("%s is processed!\n", argv[file_pos]); }
}
sort(chrvec.begin(), chrvec.end());
// Shrink and build up Refs
int curp = 0;
for (int i = 1; i <= M; ++i) {
const Transcript& transcript = transcripts.getTranscriptAt(i);
if (seqs[i] == "")
printf("Warning: Cannot extract transcript %s because the chromosome it locates -- %s -- is absent!\n", transcript.getTranscriptID().c_str(), transcript.getSeqName().c_str());
else {
refs.addRef(transcript.getTranscriptID(), seqs[i]); // insert RefSeqs
++curp;
transcripts.move(i, curp);
}
}
printf("%d transcripts are extracted and %d transcripts are omitted.\n", curp, M - curp);
transcripts.setM(curp);
M = transcripts.getM();
general_assert(M > 0, "The reference contains no transcripts!");
assert(refs.getM() == M);
}
else {
void writeToDisk(char* refName) {
ofstream fout;
sprintf(tiF, "%s.ti", refName);
transcripts.writeTo(tiF);
if (verbose) { printf("Transcript Information File is generated!\n"); }
sprintf(refFastaF, "%s.transcripts.fa", refName);
refs.writeTo(refFastaF);
sprintf(transListF, "%s.translist", refName);
refs.writeTransListTo(transListF);
sprintf(chromListF, "%s.chrlist", refName);
fout.open(chromListF);
for (int i = 0; i < (int)chrvec.size(); ++i)
fout<< chrvec[i].name<< '\t'<< chrvec[i].len<< endl;
fout.close();
if (verbose) { printf("Chromosome List File is generated!\n"); }
string cur_gene_id, cur_transcript_id, name;
vector<int> gi, gt, ta;
cur_gene_id = ""; gi.clear();
if (mappingType == 2) { cur_transcript_id = ""; gt.clear(); ta.clear(); }
for (int i = 1; i <= M; ++i) {
const Transcript& transcript = transcripts.getTranscriptAt(i);
if (cur_gene_id != transcript.getGeneID()) {
gi.push_back(i);
if (mappingType == 2) gt.push_back((int)ta.size());
cur_gene_id = transcript.getGeneID();
}
if ((mappingType == 2) && (cur_transcript_id != transcript.getTranscriptID())) {
ta.push_back(i);
cur_transcript_id = transcript.getTranscriptID();
}
}
gi.push_back(M + 1);
if (mappingType == 2) { gt.push_back((int)ta.size()); ta.push_back(M + 1); }
sprintf(groupF, "%s.grp", refName);
fout.open(groupF);
for (int i = 0; i < (int)gi.size(); ++i) fout<< gi[i]<< endl;
fout.close();
if (verbose) { printf("Group File is generated!\n"); }
if (mappingType == 2) {
sprintf(gtF, "%s.gt", refName);
fout.open(gtF);
for (int i = 0; i < (int)gt.size(); ++i) fout<< gt[i]<< endl;
fout.close();
sprintf(taF, "%s.ta", refName);
fout.open(taF);
for (int i = 0; i < (int)ta.size(); ++i) fout<< ta[i]<< endl;
fout.close();
if (verbose) { printf("Allele-specific group files are generated!\n"); }
}
if (n2g_idx) {
sprintf(n2g_idxF, "%s.n2g.idx.fa", refName);
fout.open(n2g_idxF);
for (int i = 1; i <= M; ++i)
fout<< '>'<< refs.getRef(i)->getName()<< endl<< n2g(refs.getRef(i)->getSeq())<< endl;
fout.close();
if (verbose) printf("%s is generated!\n", n2g_idxF);
}
}
int main(int argc, char* argv[]) {
if (argc < 8) {
printf("Usage: rsem-calculate-credibility-intervals reference_name sample_name sampleToken confidence nCV nSpC nMB [-p #Threads] [-q]\n");
exit(-1);
}
confidence = atof(argv[4]);
nCV = atoi(argv[5]);
nSpC = atoi(argv[6]);
nMB = atoi(argv[7]);
nThreads = 1;
quiet = false;
for (int i = 8; i < argc; i++) {
if (!strcmp(argv[i], "-p")) nThreads = atoi(argv[i + 1]);
if (!strcmp(argv[i], "-q")) quiet = true;
}
verbose = !quiet;
sprintf(refF, "%s.seq", argv[1]);
refs.loadRefs(refF, 1);
M = refs.getM();
sprintf(groupF, "%s.grp", argv[1]);
gi.load(groupF);
m = gi.getm();
nSamples = nCV * nSpC;
cvlen = M + 1;
assert(nSamples > 0 && cvlen > 1); // for Buffter.h: (bufsize_type)nSamples
sprintf(imdName, "%s.temp/%s", argv[2], argv[3]);
sprintf(statName, "%s.stat/%s", argv[2], argv[3]);
sprintf(tmpF, "%s.tmp", imdName);
sprintf(cvsF, "%s.countvectors", imdName);
sprintf(modelF, "%s.model", statName);
FILE *fi = fopen(modelF, "r");
general_assert(fi != NULL, "Cannot open " + cstrtos(modelF) + "!");
assert(fscanf(fi, "%d", &model_type) == 1);
fclose(fi);
// Phase I
switch(model_type) {
case 0 : sample_theta_vectors_from_count_vectors<SingleModel>(); break;
case 1 : sample_theta_vectors_from_count_vectors<SingleQModel>(); break;
case 2 : sample_theta_vectors_from_count_vectors<PairedEndModel>(); break;
case 3 : sample_theta_vectors_from_count_vectors<PairedEndQModel>(); break;
}
// Phase II
calculate_credibility_intervals(imdName);
/*
sprintf(command, "rm -f %s", tmpF);
int status = system(command);
if (status != 0) {
fprintf(stderr, "Cannot delete %s!\n", tmpF);
exit(-1);
}
*/
return 0;
}
int main(int argc, char* argv[]) {
ifstream fin;
bool quiet = false;
if (argc < 6) {
printf("Usage : rsem-run-em refName read_type sampleName imdName statName [-p #Threads] [-b samInpType samInpF has_fn_list_? [fn_list]] [-q] [--gibbs-out] [--sampling] [--seed seed] [--calc-evaluation-score nb_r nb_p L w]\n\n");
printf(" refName: reference name\n");
printf(" read_type: 0 single read without quality score; 1 single read with quality score; 2 paired-end read without quality score; 3 paired-end read with quality score.\n");
printf(" sampleName: sample's name, including the path\n");
printf(" sampleToken: sampleName excludes the path\n");
printf(" -p: number of threads which user wants to use. (default: 1)\n");
printf(" -b: produce bam format output file. (default: off)\n");
printf(" -q: set it quiet\n");
printf(" --gibbs-out: generate output file use by Gibbs sampler. (default: off)\n");
printf(" --sampling: sample each read from its posterior distribution when bam file is generated. (default: off)\n");
printf(" --seed uint32: the seed used for the BAM sampling. (default: off)\n");
printf(" --calc-evaluation-score nb_r nb_p L w: "
"nb_r and nb_p are parameters for the true transcript length distribution, which is modeled by a negative binomial distribution; "
"L is the read length and w is the mininum overlap required for joining two contigs.\n");
printf("// model parameters should be in imdName.mparams.\n");
exit(-1);
}
time_t a = time(NULL);
strcpy(refName, argv[1]);
read_type = atoi(argv[2]);
strcpy(outName, argv[3]);
strcpy(imdName, argv[4]);
strcpy(statName, argv[5]);
nThreads = 1;
genBamF = false;
bamSampling = false;
genGibbsOut = false;
calcEvalScore = false;
pt_fn_list = NULL;
hasSeed = false;
for (int i = 6; i < argc; i++) {
if (!strcmp(argv[i], "-p")) { nThreads = atoi(argv[i + 1]); }
if (!strcmp(argv[i], "-b")) {
genBamF = true;
inpSamType = argv[i + 1][0];
strcpy(inpSamF, argv[i + 2]);
if (atoi(argv[i + 3]) == 1) {
strcpy(fn_list, argv[i + 4]);
pt_fn_list = (char*)(&fn_list);
}
}
if (!strcmp(argv[i], "-q")) { quiet = true; }
if (!strcmp(argv[i], "--gibbs-out")) { genGibbsOut = true; }
if (!strcmp(argv[i], "--sampling")) { bamSampling = true; }
if (!strcmp(argv[i], "--seed")) {
hasSeed = true;
int len = strlen(argv[i + 1]);
seed = 0;
for (int k = 0; k < len; k++) seed = seed * 10 + (argv[i + 1][k] - '0');
}
if (!strcmp(argv[i], "--calc-evaluation-score")) {
calcEvalScore = true;
nb_r = atof(argv[i + 1]);
nb_p = atof(argv[i + 2]);
L = atoi(argv[i + 3]);
w = atoi(argv[i + 4]);
}
}
general_assert(nThreads > 0, "Number of threads should be bigger than 0!");
verbose = !quiet;
//basic info loading
sprintf(refF, "%s.seq", refName);
refs.loadRefs(refF);
M = refs.getM();
sprintf(tiF, "%s.ti", refName);
transcripts.readFrom(tiF);
sprintf(cntF, "%s.cnt", statName);
fin.open(cntF);
general_assert(fin.is_open(), "Cannot open " + cstrtos(cntF) + "! It may not exist.");
fin>>N0>>N1>>N2>>N_tot;
fin.close();
general_assert(N1 > 0, "There are no alignable reads!");
if ((READ_INT_TYPE)nThreads > N1) nThreads = N1;
//set model parameters
mparams.M = M;
mparams.N[0] = N0; mparams.N[1] = N1; mparams.N[2] = N2;
mparams.refs = &refs;
sprintf(mparamsF, "%s.mparams", imdName);
fin.open(mparamsF);
general_assert(fin.is_open(), "Cannot open " + cstrtos(mparamsF) + "It may not exist.");
fin>> mparams.minL>> mparams.maxL>> mparams.probF;
int val; // 0 or 1 , for estRSPD
fin>>val;
mparams.estRSPD = (val != 0);
fin>> mparams.B>> mparams.mate_minL>> mparams.mate_maxL>> mparams.mean>> mparams.sd;
fin>> mparams.seedLen;
fin.close();
//run EM
switch(read_type) {
case 0 : EM<SingleRead, SingleHit, SingleModel>(); break;
case 1 : EM<SingleReadQ, SingleHit, SingleQModel>(); break;
case 2 : EM<PairedEndRead, PairedEndHit, PairedEndModel>(); break;
case 3 : EM<PairedEndReadQ, PairedEndHit, PairedEndQModel>(); break;
default : fprintf(stderr, "Unknown Read Type!\n"); exit(-1);
}
if (calcEvalScore) {
CalcEvalScore ces(refs, nb_r, nb_p, L, w, statName);
sprintf(scoreF, "%s.score", outName);
ces.writeScoresTo(scoreF);
char groupF[STRLEN];
GroupInfo gi;
sprintf(groupF, "%s.grp", argv[1]);
gi.load(groupF);
ces.generateExpressionFiles(gi, transcripts, scoreF);
}
time_t b = time(NULL);
printTimeUsed(a, b, "EM.cpp");
return 0;
}