int splicing_i_gff_constitutive_exons_all(const splicing_gff_t *gff,
splicing_exonset_t *exons,
int min_length) {
size_t g, nogenes;
splicing_vector_int_t events;
SPLICING_CHECK(splicing_gff_nogenes(gff, &nogenes));
SPLICING_CHECK(splicing_exonset_init(exons, 0));
SPLICING_FINALLY(splicing_exonset_destroy, exons);
SPLICING_CHECK(splicing_vector_int_init(&events, 0));
SPLICING_FINALLY(splicing_vector_int_destroy, &events);
for (g=0; g<nogenes; g++) {
const char *seqid=
splicing_strvector_get(&gff->seqids, VECTOR(gff->seqid)[g]);
int noex, idx, noEvents, i;
size_t noiso;
int start=VECTOR(gff->genes)[g];
int end= g+1 < nogenes ? VECTOR(gff->genes)[g+1] : gff->n;
splicing_gff_noiso_one(gff, g, &noiso);
/* Collect and sort all events */
splicing_vector_int_clear(&events);
for (idx=start+1; idx<end; idx++) {
if (VECTOR(gff->type)[idx] == SPLICING_TYPE_EXON) {
SPLICING_CHECK(splicing_vector_int_push_back2
(&events, VECTOR(gff->start)[idx],
-VECTOR(gff->end)[idx]));
}
}
noEvents=splicing_vector_int_size(&events);
splicing_qsort(VECTOR(events), noEvents, sizeof(int),
splicing_i_const_cmp);
/* Now go over the sorted events and extract the constitutive exons */
for (noex=0, i=0; i<noEvents; i++) {
int ev=VECTOR(events)[i];
if (ev > 0) { noex++; }
if (ev < 0) {
int prev=VECTOR(events)[i-1];
if (noex == noiso && (-ev)-prev+1 >= min_length) {
/* constitutive exon */
SPLICING_CHECK(splicing_exonset_append(exons, seqid, prev, -ev));
}
noex--;
}
}
}
splicing_vector_int_destroy(&events);
SPLICING_FINALLY_CLEAN(2); /* + exons */
return 0;
}
int splicing_genomic_to_iso(const splicing_gff_t *gff, size_t gene,
const splicing_vector_int_t *position,
splicing_matrix_int_t *isopos) {
size_t r, i, noiso, noreads=splicing_vector_int_size(position);
splicing_vector_int_t exstart, exend, exidx, shift;
splicing_gff_noiso_one(gff, gene, &noiso);
SPLICING_CHECK(splicing_vector_int_init(&exstart, 0));
SPLICING_FINALLY(splicing_vector_int_destroy, &exstart);
SPLICING_CHECK(splicing_vector_int_init(&exend, 0));
SPLICING_FINALLY(splicing_vector_int_destroy, &exend);
SPLICING_CHECK(splicing_vector_int_init(&exidx, 0));
SPLICING_FINALLY(splicing_vector_int_destroy, &exidx);
SPLICING_CHECK(splicing_gff_exon_start_end(gff, &exstart, &exend,
&exidx, gene));
SPLICING_CHECK(splicing_vector_int_init(&shift, 0));
SPLICING_FINALLY(splicing_vector_int_destroy, &shift);
for (i=0; i<noiso; i++) {
size_t cs=0, ce=0, ex=0;
int pos=VECTOR(exidx)[i], pos2=VECTOR(exidx)[i+1];
while (pos < pos2) {
cs += VECTOR(exstart)[pos];
SPLICING_CHECK(splicing_vector_int_push_back(&shift, cs-ce-ex-1));
ex++; ce += VECTOR(exend)[pos]; pos++;
}
}
SPLICING_CHECK(splicing_matrix_int_resize(isopos, noiso, noreads));
for (r=0; r<noreads; r++) {
for (i=0; i<noiso; i++) {
size_t pos=VECTOR(*position)[r];
size_t startpos=VECTOR(exidx)[i];
size_t endpos=VECTOR(exidx)[i+1];
int ex;
for (ex=startpos; ex < endpos && VECTOR(exend)[ex] < pos; ex++) ;
if (VECTOR(exstart)[ex] <= pos && pos <= VECTOR(exend)[ex]) {
MATRIX(*isopos, i, r) = VECTOR(*position)[r] - VECTOR(shift)[ex];
} else {
MATRIX(*isopos, i, r) = -1;
}
}
}
splicing_vector_int_destroy(&shift);
splicing_vector_int_destroy(&exidx);
splicing_vector_int_destroy(&exend);
splicing_vector_int_destroy(&exstart);
SPLICING_FINALLY_CLEAN(4);
return 0;
}
int splicing_gff_exon_start_end(const splicing_gff_t *gff,
splicing_vector_int_t *start,
splicing_vector_int_t *end,
splicing_vector_int_t *idx,
int gene) {
size_t noiso;
int i=0, p=0, n=splicing_gff_size(gff);
int pos;
size_t nogenes;
splicing_vector_int_t tmp, tmp2;
SPLICING_CHECK(splicing_vector_int_init(&tmp, 10));
SPLICING_FINALLY(splicing_vector_int_destroy, &tmp);
SPLICING_CHECK(splicing_vector_int_init(&tmp2, 10));
SPLICING_FINALLY(splicing_vector_int_destroy, &tmp2);
SPLICING_CHECK(splicing_gff_nogenes(gff, &nogenes));
if (gene < 0 || gene >= nogenes) {
SPLICING_ERROR("Invalid gene id", SPLICING_EINVAL);
}
pos=VECTOR(gff->genes)[gene]+1;
SPLICING_CHECK(splicing_gff_noiso_one(gff, gene, &noiso));
splicing_vector_int_clear(start);
splicing_vector_int_clear(end);
SPLICING_CHECK(splicing_vector_int_resize(idx, noiso+1));
while (pos < n) {
if (VECTOR(gff->type)[pos] == SPLICING_TYPE_EXON) {
int s=VECTOR(gff->start)[pos];
int e=VECTOR(gff->end)[pos];
SPLICING_CHECK(splicing_vector_int_push_back(start, s)); p++;
SPLICING_CHECK(splicing_vector_int_push_back(end, e));
} else if (VECTOR(gff->type)[pos] == SPLICING_TYPE_MRNA) {
VECTOR(*idx)[i] = p;
if (i!=0) {
SPLICING_CHECK(splicing_i_gff_exon_start_end_sort(start, end, idx,
i-1, &tmp, &tmp2));
}
i++;
} else if (VECTOR(gff->type)[pos] == SPLICING_TYPE_GENE) {
break;
}
pos++;
}
VECTOR(*idx)[i] = p;
SPLICING_CHECK(splicing_i_gff_exon_start_end_sort(start, end, idx, i-1,
&tmp, &tmp2));
splicing_vector_int_destroy(&tmp2);
splicing_vector_int_destroy(&tmp);
SPLICING_FINALLY_CLEAN(1);
return 0;
}
int splicing_i_gff_constitutive_exons_full(const splicing_gff_t *gff,
splicing_exonset_t *exons,
int min_length) {
size_t g, nogenes;
splicing_vector_int_t events;
SPLICING_CHECK(splicing_gff_nogenes(gff, &nogenes));
SPLICING_CHECK(splicing_exonset_init(exons, 1));
SPLICING_FINALLY(splicing_exonset_destroy, exons);
SPLICING_CHECK(splicing_vector_int_init(&events, 0));
SPLICING_FINALLY(splicing_vector_int_destroy, &events);
for (g=0; g<nogenes; g++) {
const char *seqid=
splicing_strvector_get(&gff->seqids, VECTOR(gff->seqid)[g]);
int i, idx, noExons, noSame;
size_t noiso;
int start=VECTOR(gff->genes)[g];
int end= g+1 < nogenes ? VECTOR(gff->genes)[g+1] : gff->n;
splicing_gff_noiso_one(gff, g, &noiso);
/* Collect and sort all events */
splicing_vector_int_clear(&events);
for (idx=start+1; idx<end; idx++) {
if (VECTOR(gff->type)[idx] == SPLICING_TYPE_EXON) {
SPLICING_CHECK(splicing_vector_int_push_back2
(&events, VECTOR(gff->start)[idx],
VECTOR(gff->end)[idx]));
}
}
noExons=splicing_vector_int_size(&events)/2;
splicing_qsort(VECTOR(events), noExons, sizeof(int)*2,
splicing_i_const_cmp2);
/* Now go over them and check how many times each exon appears */
for (noSame=1, i=2; i<noExons*2; i+=2) {
int start=VECTOR(events)[i];
int end=VECTOR(events)[i+1];
if (start == VECTOR(events)[i-2] && VECTOR(events)[i-1] == end) {
noSame++;
} else {
noSame=1;
}
if (noSame == noiso) {
SPLICING_CHECK(splicing_exonset_append(exons, seqid, start, end));
}
}
}
splicing_vector_int_destroy(&events);
SPLICING_FINALLY_CLEAN(2); /* + exons */
return 0;
}
int splicing_gff_converter_init(const splicing_gff_t *gff, size_t gene,
splicing_gff_converter_t *converter) {
int i;
SPLICING_CHECK(splicing_gff_noiso_one(gff, gene, &converter->noiso));
SPLICING_VECTOR_INT_INIT_FINALLY(&converter->exstart, 0);
SPLICING_VECTOR_INT_INIT_FINALLY(&converter->exend, 0);
SPLICING_VECTOR_INT_INIT_FINALLY(&converter->exidx, 0);
SPLICING_VECTOR_INT_INIT_FINALLY(&converter->shift, 0);
SPLICING_VECTOR_INT_INIT_FINALLY(&converter->exlim, 0);
SPLICING_CHECK(splicing_gff_exon_start_end(gff, &converter->exstart,
&converter->exend,
&converter->exidx, gene));
/* Calculate the shift */
for (i=0; i < converter->noiso; i++) {
size_t cs=0, ce=0, ex=0;
int pos=VECTOR(converter->exidx)[i], pos2=VECTOR(converter->exidx)[i+1];
while (pos < pos2) {
cs += VECTOR(converter->exstart)[pos];
SPLICING_CHECK(splicing_vector_int_push_back(&converter->shift,
cs-ce-ex-1));
ex++; ce += VECTOR(converter->exend)[pos]; pos++;
}
}
/* Calculate the exlim */
for (i=0; i < converter->noiso; i++) {
size_t cs=0;
int pos=VECTOR(converter->exidx)[i], pos2=VECTOR(converter->exidx)[i+1];
while (pos < pos2) {
size_t l=
VECTOR(converter->exend)[pos] - VECTOR(converter->exstart)[pos]+1;
cs += l;
SPLICING_CHECK(splicing_vector_int_push_back(&converter->exlim, cs+1));
pos++;
}
}
SPLICING_FINALLY_CLEAN(5);
return 0;
}
int splicing_miso(const splicing_gff_t *gff, size_t gene,
const splicing_vector_int_t *position,
const char **cigarstr, int readLength,
int noIterations, int noBurnIn, int noLag,
const splicing_vector_t *hyperp,
splicing_matrix_t *samples, splicing_vector_t *logLik,
splicing_matrix_t *match_matrix,
splicing_matrix_t *class_templates,
splicing_vector_t *class_counts,
splicing_vector_int_t *assignment,
splicing_miso_rundata_t *rundata) {
double acceptP, cJS, pJS, sigma;
int noReads = splicing_vector_int_size(position);
splicing_vector_int_t *myass=assignment, vass;
size_t noiso;
splicing_vector_t vpsi, vpsiNew, valpha, valphaNew,
*psi=&vpsi, *psiNew=&vpsiNew, *alpha=&valpha, *alphaNew=&valphaNew;
int noSamples = (noIterations - noBurnIn + 1) / noLag;
int i, m, lagCounter=0, noS=0;
splicing_matrix_t *mymatch_matrix=match_matrix, vmatch_matrix;
splicing_vector_int_t match_order;
splicing_vector_int_t effisolen;
splicing_vector_t isoscores;
if ( (class_templates ? 1 : 0) + (class_counts ? 1 : 0) == 1) {
SPLICING_ERROR("Only one of `class_templates' and `class_counts' is "
"given", SPLICING_EINVAL);
}
SPLICING_CHECK(splicing_gff_noiso_one(gff, gene, &noiso));
rundata->noIso=noiso;
rundata->noIters=noIterations;
rundata->noBurnIn=noBurnIn;
rundata->noLag=noLag;
rundata->noAccepted = rundata->noRejected = 0;
if (assignment) {
SPLICING_CHECK(splicing_vector_int_resize(myass, noReads));
splicing_vector_int_null(myass);
} else {
myass=&vass;
SPLICING_CHECK(splicing_vector_int_init(myass, noReads));
SPLICING_FINALLY(splicing_vector_int_destroy, myass);
}
SPLICING_CHECK(splicing_vector_init(&vpsi, noiso));
SPLICING_FINALLY(splicing_vector_destroy, &vpsi);
SPLICING_CHECK(splicing_vector_init(&vpsiNew, noiso));
SPLICING_FINALLY(splicing_vector_destroy, &vpsiNew);
SPLICING_CHECK(splicing_vector_init(&valpha, noiso-1));
SPLICING_FINALLY(splicing_vector_destroy, &valpha);
SPLICING_CHECK(splicing_vector_init(&valphaNew, noiso-1));
SPLICING_FINALLY(splicing_vector_destroy, &valphaNew);
if (match_matrix) {
SPLICING_CHECK(splicing_matrix_resize(match_matrix, noiso, noReads));
} else {
mymatch_matrix=&vmatch_matrix;
SPLICING_CHECK(splicing_matrix_init(mymatch_matrix, noiso, noReads));
SPLICING_FINALLY(splicing_matrix_destroy, mymatch_matrix);
}
SPLICING_CHECK(splicing_vector_int_init(&match_order, noReads));
SPLICING_FINALLY(splicing_vector_int_destroy, &match_order);
SPLICING_CHECK(splicing_matchIso(gff, gene, position, cigarstr,
mymatch_matrix));
SPLICING_CHECK(splicing_order_matches(mymatch_matrix, &match_order));
if (class_templates && class_counts) {
SPLICING_CHECK(splicing_i_miso_classes(mymatch_matrix, &match_order,
class_templates, class_counts,
/*bin_class_templates=*/ 0,
/*bin_class_counts=*/ 0));
}
SPLICING_CHECK(splicing_vector_int_init(&effisolen, noiso));
SPLICING_FINALLY(splicing_vector_int_destroy, &effisolen);
SPLICING_CHECK(splicing_vector_init(&isoscores, noiso));
SPLICING_FINALLY(splicing_vector_destroy, &isoscores);
SPLICING_CHECK(splicing_gff_isolength_one(gff, gene, &effisolen));
for (i=0; i<noiso; i++) {
int l=VECTOR(effisolen)[i]-readLength+1;
VECTOR(effisolen)[i] = l > 0 ? l : 0;
VECTOR(isoscores)[i] = -log((double) l);
}
SPLICING_CHECK(splicing_matrix_resize(samples, noiso, noSamples));
SPLICING_CHECK(splicing_vector_resize(logLik, noSamples));
/* Initialize Psi(0) randomly */
SPLICING_CHECK(splicing_drift_proposal(/* mode= */ 0, 0, 0, 0, 0, 0,
noiso, psi, alpha, &sigma, 0));
SPLICING_CHECK(splicing_drift_proposal(/* mode= */ 1, psi, alpha, sigma,
0, 0, noiso, psi, alpha, 0, 0));
/* Initialize assignments of reads */
SPLICING_CHECK(splicing_reassign_samples(mymatch_matrix, &match_order, psi,
noiso, myass));
/* foreach Iteration m=1, ..., M do */
for (m=0; m < noIterations; m++) {
SPLICING_CHECK(splicing_drift_proposal(/* mode= */ 1, psi, alpha, sigma,
0, 0, noiso, psiNew, alphaNew, 0,
0));
SPLICING_CHECK(splicing_metropolis_hastings_ratio(myass, noReads, psiNew,
alphaNew, psi, alpha,
sigma, noiso,
&effisolen, hyperp,
&isoscores,
m > 0 ? 1 : 0,
&acceptP, &cJS, &pJS));
if (acceptP >= 1 || RNG_UNIF01() < acceptP) {
splicing_vector_t *tmp;
tmp=psi; psi=psiNew; psiNew=tmp;
tmp=alpha; alpha=alphaNew; alphaNew=tmp;
cJS = pJS;
rundata->noAccepted ++;
} else {
rundata->noRejected ++;
}
if (m >= noBurnIn) {
if (lagCounter == noLag - 1) {
memcpy(&MATRIX(*samples, 0, noS), VECTOR(*psi),
noiso * sizeof(double));
VECTOR(*logLik)[noS] = cJS;
noS++;
lagCounter = 0;
} else {
lagCounter ++;
}
}
SPLICING_CHECK(splicing_reassign_samples(mymatch_matrix, &match_order,
psi, noiso, myass));
} /* for m < noIterations */
splicing_vector_destroy(&isoscores);
splicing_vector_int_destroy(&effisolen);
splicing_vector_int_destroy(&match_order);
SPLICING_FINALLY_CLEAN(3);
if (!match_matrix) {
splicing_matrix_destroy(mymatch_matrix);
SPLICING_FINALLY_CLEAN(1);
}
splicing_vector_destroy(&valphaNew);
splicing_vector_destroy(&valpha);
splicing_vector_destroy(&vpsiNew);
splicing_vector_destroy(&vpsi);
SPLICING_FINALLY_CLEAN(4);
if (!assignment) {
splicing_vector_int_destroy(myass);
SPLICING_FINALLY_CLEAN(1);
}
return 0;
}
int splicing_iso_to_genomic(const splicing_gff_t *gff, size_t gene,
const splicing_vector_int_t *isoform,
const splicing_vector_int_t *exstart,
const splicing_vector_int_t *exend,
const splicing_vector_int_t *exidx,
splicing_vector_int_t *position) {
size_t i, noiso, n=splicing_vector_int_size(position);
splicing_vector_int_t exlim, shift;
splicing_vector_int_t vexstart, vexend, vexidx,
*myexstart=(splicing_vector_int_t *) exstart,
*myexend=(splicing_vector_int_t *) exend,
*myexidx=(splicing_vector_int_t *) exidx;
size_t pos, pos2;
if (!exstart || !exend || !exidx) {
myexstart=&vexstart;
myexend=&vexend;
myexidx=&vexidx;
SPLICING_CHECK(splicing_vector_int_init(myexstart, 0));
SPLICING_FINALLY(splicing_vector_int_destroy, myexstart);
SPLICING_CHECK(splicing_vector_int_init(myexend, 0));
SPLICING_FINALLY(splicing_vector_int_destroy, myexend);
SPLICING_CHECK(splicing_vector_int_init(myexidx, 0));
SPLICING_FINALLY(splicing_vector_int_destroy, myexidx);
SPLICING_CHECK(splicing_gff_exon_start_end(gff, myexstart, myexend,
myexidx, gene));
}
SPLICING_CHECK(splicing_gff_noiso_one(gff, gene, &noiso));
SPLICING_CHECK(splicing_vector_int_init(&exlim, 0));
SPLICING_FINALLY(splicing_vector_int_destroy, &exlim);
SPLICING_CHECK(splicing_vector_int_init(&shift, 0));
SPLICING_FINALLY(splicing_vector_int_destroy, &shift);
for (i=0; i<noiso; i++) {
size_t cs=0, ce=0, ex=0;
int pos=VECTOR(*myexidx)[i], pos2=VECTOR(*myexidx)[i+1];
while (pos < pos2) {
cs += VECTOR(*myexstart)[pos];
SPLICING_CHECK(splicing_vector_int_push_back(&shift, cs-ce-ex-1));
ex++; ce += VECTOR(*myexend)[pos]; pos++;
}
}
for (i=0; i<noiso; i++) {
size_t cs=0;
int pos=VECTOR(*myexidx)[i], pos2=VECTOR(*myexidx)[i+1];
while (pos < pos2) {
size_t l=VECTOR(*myexend)[pos]-VECTOR(*myexstart)[pos]+1;
cs += l;
SPLICING_CHECK(splicing_vector_int_push_back(&exlim, cs+1));
pos++;
}
}
for (i=0; i<n; i++) {
int iso=VECTOR(*isoform)[i];
size_t pos=VECTOR(*position)[i];
int ex;
for (ex=VECTOR(*myexidx)[iso]; VECTOR(exlim)[ex] <= pos; ex++) ;
VECTOR(*position)[i] = pos + VECTOR(shift)[ex];
}
splicing_vector_int_destroy(&shift);
splicing_vector_int_destroy(&exlim);
SPLICING_FINALLY_CLEAN(2);
if (!exstart || !exend || !exidx) {
splicing_vector_int_destroy(myexidx);
splicing_vector_int_destroy(myexend);
splicing_vector_int_destroy(myexstart);
SPLICING_FINALLY_CLEAN(3);
}
return 0;
}
static PyObject* pysplicing_miso(PyObject *self, PyObject *args) {
PyObject *gff, *readpos, *readcigar, *hyperp=0;
int gene, readLength, noIterations=5000, noBurnIn=500, noLag=10;
splicing_gff_t *mygff;
splicing_strvector_t myreadcigar;
splicing_vector_int_t myreadpos;
splicing_vector_t myhyperp;
splicing_matrix_t samples;
splicing_vector_t logLik;
splicing_matrix_t class_templates;
splicing_vector_t class_counts;
splicing_vector_int_t assignment;
splicing_miso_rundata_t rundata;
PyObject *r1, *r2, *r3, *r4, *r5, *r6;
if (!PyArg_ParseTuple(args, "OiOOi|iiiO", &gff, &gene, &readpos, &readcigar,
&readLength, &noIterations, &noBurnIn, &noLag,
&hyperp)) { return NULL; }
mygff=PyCObject_AsVoidPtr(gff);
SPLICING_PYCHECK(splicing_matrix_init(&samples, 0, 0));
SPLICING_FINALLY(splicing_matrix_destroy, &samples);
SPLICING_PYCHECK(splicing_vector_init(&logLik, 0));
SPLICING_FINALLY(splicing_vector_destroy, &logLik);
SPLICING_PYCHECK(splicing_matrix_init(&class_templates, 0, 0));
SPLICING_FINALLY(splicing_matrix_destroy, &class_templates);
SPLICING_PYCHECK(splicing_vector_init(&class_counts, 0));
SPLICING_FINALLY(splicing_vector_destroy, &class_counts);
SPLICING_PYCHECK(splicing_vector_int_init(&assignment, 0));
SPLICING_FINALLY(splicing_vector_int_destroy, &assignment);
if (pysplicing_to_vector_int(readpos, &myreadpos)) { return NULL; }
SPLICING_FINALLY(splicing_vector_int_destroy, &myreadpos);
if (hyperp) {
if (pysplicing_to_vector(hyperp, &myhyperp)) { return NULL; }
SPLICING_FINALLY(splicing_vector_destroy, &myhyperp);
} else {
size_t i, noiso;
SPLICING_PYCHECK(splicing_gff_noiso_one(mygff, gene, &noiso));
SPLICING_PYCHECK(splicing_vector_init(&myhyperp, noiso));
SPLICING_FINALLY(splicing_vector_destroy, &myhyperp);
for (i=0; i<noiso; i++) { VECTOR(myhyperp)[i] = 1.0; }
}
if (pysplicing_to_strvector(readcigar, &myreadcigar)) { return NULL; };
SPLICING_FINALLY(splicing_strvector_destroy, &myreadcigar);
SPLICING_PYCHECK(splicing_miso(mygff, gene, &myreadpos,
(const char**) myreadcigar.table,
readLength, noIterations, noBurnIn, noLag,
&myhyperp, &samples, &logLik,
/*match_matrix=*/ 0, &class_templates,
&class_counts, &assignment, &rundata));
splicing_vector_destroy(&myhyperp);
splicing_vector_int_destroy(&myreadpos);
splicing_strvector_destroy(&myreadcigar);
SPLICING_FINALLY_CLEAN(3);
r6=pysplicing_from_miso_rundata(&rundata);
r5=pysplicing_from_vector_int(&assignment);
splicing_vector_int_destroy(&assignment); SPLICING_FINALLY_CLEAN(1);
r4=pysplicing_from_vector(&class_counts);
splicing_vector_destroy(&class_counts); SPLICING_FINALLY_CLEAN(1);
splicing_matrix_transpose(&class_templates);
r3=pysplicing_from_matrix(&class_templates);
splicing_matrix_destroy(&class_templates); SPLICING_FINALLY_CLEAN(1);
r2=pysplicing_from_vector(&logLik);
splicing_vector_destroy(&logLik); SPLICING_FINALLY_CLEAN(1);
r1=pysplicing_from_matrix(&samples);
splicing_matrix_destroy(&samples); SPLICING_FINALLY_CLEAN(1);
return Py_BuildValue("OOOOOO", r1, r2, r3, r4, r5, r6);
}
int splicing_simulate_reads(const splicing_gff_t *gff, int gene,
const splicing_vector_t *expression,
int noreads, int readLength,
splicing_vector_int_t *isoform,
splicing_vector_int_t *position,
splicing_strvector_t *cigar,
splicing_vector_t *sample_prob) {
size_t i, p, noiso, goodiso=0, nogenes;
splicing_vector_int_t effisolen;
splicing_vector_t sampleprob;
double rand, sumpsi=0.0;
splicing_vector_int_t exstart, exend, exidx;
SPLICING_CHECK(splicing_gff_nogenes(gff, &nogenes));
if (gene < 0 || gene >= nogenes) {
SPLICING_ERROR("Invalid gene id", SPLICING_EINVAL);
}
/* TODO: more error checks */
SPLICING_CHECK(splicing_gff_noiso_one(gff, gene, &noiso));
SPLICING_CHECK(splicing_vector_int_init(&effisolen, noiso));
SPLICING_FINALLY(splicing_vector_int_destroy, &effisolen);
SPLICING_CHECK(splicing_vector_init(&sampleprob, noiso));
SPLICING_FINALLY(splicing_vector_destroy, &sampleprob);
SPLICING_CHECK(splicing_vector_int_resize(isoform, noreads));
SPLICING_CHECK(splicing_gff_isolength_one(gff, gene, &effisolen));
for (i=0; i<noiso; i++) {
int l=VECTOR(effisolen)[i]-readLength+1;
VECTOR(effisolen)[i] = l > 0 ? l : 0;
VECTOR(sampleprob)[i] = VECTOR(*expression)[i] * VECTOR(effisolen)[i];
if (VECTOR(sampleprob)[i] != 0) { goodiso++; }
sumpsi += VECTOR(sampleprob)[i];
}
if (goodiso==0) {
SPLICING_ERROR("No isoform is possible", SPLICING_FAILURE);
}
if (sample_prob) {
SPLICING_CHECK(splicing_vector_update(sample_prob, &sampleprob));
}
for (i=1; i<noiso; i++) {
VECTOR(sampleprob)[i] += VECTOR(sampleprob)[i-1];
}
for (i=0; i<noreads; i++) {
int w;
if (noiso==1) {
w=0;
} else if (noiso==2) {
rand = RNG_UNIF01() * sumpsi;
w = (rand < VECTOR(sampleprob)[0]) ? 0 : 1;
} else {
rand = RNG_UNIF01() * sumpsi;
for (w=0; rand > VECTOR(sampleprob)[w]; w++) ;
}
VECTOR(*isoform)[i]=w;
}
splicing_vector_destroy(&sampleprob);
SPLICING_FINALLY_CLEAN(1);
/* OK, we have the isoforms, now we need the read positions,
these are uniformly sampled from the individual isoforms. */
SPLICING_CHECK(splicing_vector_int_resize(position, noreads));
SPLICING_CHECK(splicing_vector_int_init(&exstart, 0));
SPLICING_FINALLY(splicing_vector_int_destroy, &exstart);
SPLICING_CHECK(splicing_vector_int_init(&exend, 0));
SPLICING_FINALLY(splicing_vector_int_destroy, &exend);
SPLICING_CHECK(splicing_vector_int_init(&exidx, 0));
SPLICING_FINALLY(splicing_vector_int_destroy, &exidx);
SPLICING_CHECK(splicing_gff_exon_start_end(gff, &exstart, &exend, &exidx,
gene));
/* Positions in isoform coordinates first */
for (i=0; i<noreads; i++) {
int iso=VECTOR(*isoform)[i];
int len=VECTOR(effisolen)[iso];
VECTOR(*position)[i]=RNG_INTEGER(1, len);
}
/* Translate isoform coordinates to genomic coordintes */
/* TODO: some of this is already calculated */
SPLICING_CHECK(splicing_iso_to_genomic(gff, gene, isoform, /*converter=*/ 0,
position));
/* CIGAR strings */
splicing_strvector_clear(cigar);
SPLICING_CHECK(splicing_strvector_reserve(cigar, noreads));
for (i=0; i<noreads; i++) {
char tmp[1000], *tmp2=tmp;
int iso=VECTOR(*isoform)[i];
size_t rs=VECTOR(*position)[i];
int ex=0;
int rl=readLength;
for (ex=VECTOR(exidx)[iso]; VECTOR(exend)[ex] < rs; ex++) ;
while (VECTOR(exend)[ex] < rs+rl-1) {
tmp2 += snprintf(tmp2, sizeof(tmp)/sizeof(char)-(tmp2-tmp)-1, "%iM%iN",
(int) (VECTOR(exend)[ex]-rs+1),
(int) (VECTOR(exstart)[ex+1]-VECTOR(exend)[ex]-1));
if (tmp2 >= tmp + sizeof(tmp)/sizeof(char)) {
SPLICING_ERROR("CIGAR string too long", SPLICING_EINVAL);
}
rl -= (VECTOR(exend)[ex] - rs + 1);
rs = VECTOR(exstart)[ex+1];
ex++;
}
tmp2 += snprintf(tmp2, sizeof(tmp)/sizeof(char)-(tmp2-tmp)-1, "%iM", rl);
if (tmp2 >= tmp + sizeof(tmp)/sizeof(char)) {
SPLICING_ERROR("CIGAR string too long", SPLICING_EINVAL); }
SPLICING_CHECK(splicing_strvector_append(cigar, tmp));
}
splicing_vector_int_destroy(&exidx);
splicing_vector_int_destroy(&exend);
splicing_vector_int_destroy(&exstart);
splicing_vector_int_destroy(&effisolen);
SPLICING_FINALLY_CLEAN(4);
return 0;
}
int splicing_simulate_paired_reads(const splicing_gff_t *gff, int gene,
const splicing_vector_t *expression,
int noreads, int readLength,
const splicing_vector_t *fragmentProb,
int fragmentStart, double normalMean,
double normalVar, double numDevs,
splicing_vector_int_t *isoform,
splicing_vector_int_t *position,
splicing_strvector_t *cigar,
splicing_vector_t *sampleprob) {
size_t i, j, noiso, il, nogenes;
splicing_vector_t *mysampleprob=sampleprob, vsampleprob;
splicing_vector_t px, cpx;
double sumpx, sumpsi=0.0;
splicing_vector_int_t isolen;
int goodiso=0;
splicing_vector_int_t exstart, exend, exidx;
splicing_vector_t *myfragmentProb=(splicing_vector_t*) fragmentProb,
vfragmentProb;
int fs, fl;
SPLICING_CHECK(splicing_gff_nogenes(gff, &nogenes));
if (gene < 0 || gene >= nogenes) {
SPLICING_ERROR("Invalid gene id", SPLICING_EINVAL);
}
/* TODO: more error checks */
if (!fragmentProb) {
myfragmentProb=&vfragmentProb;
SPLICING_CHECK(splicing_vector_init(&vfragmentProb, 0));
SPLICING_FINALLY(splicing_vector_destroy, &vfragmentProb);
SPLICING_CHECK(splicing_normal_fragment(normalMean, normalVar, numDevs,
readLength, myfragmentProb,
&fragmentStart));
splicing_vector_scale(myfragmentProb,
1.0/splicing_vector_sum(myfragmentProb));
}
il=splicing_vector_size(myfragmentProb);
fs=fragmentStart;
fl=fragmentStart+il-1;
SPLICING_CHECK(splicing_gff_noiso_one(gff, gene, &noiso));
if ( fabs(splicing_vector_sum(myfragmentProb) - 1.0) > 1e-10 ) {
SPLICING_ERROR("Fragment length distribution does not sum up to 1",
SPLICING_EINVAL);
}
SPLICING_CHECK(splicing_vector_int_init(&isolen, noiso));
SPLICING_FINALLY(splicing_vector_int_destroy, &isolen);
SPLICING_CHECK(splicing_gff_isolength_one(gff, gene, &isolen));
SPLICING_CHECK(splicing_vector_copy(&px, myfragmentProb));
SPLICING_FINALLY(splicing_vector_destroy, &px);
SPLICING_CHECK(splicing_vector_init(&cpx, il));
SPLICING_FINALLY(splicing_vector_destroy, &cpx);
if (!sampleprob) {
mysampleprob=&vsampleprob;
SPLICING_CHECK(splicing_vector_init(mysampleprob, noiso));
SPLICING_FINALLY(splicing_vector_destroy, mysampleprob);
} else {
SPLICING_CHECK(splicing_vector_resize(mysampleprob, noiso));
}
for (sumpx=VECTOR(px)[0], i=1; i<il; i++) {
VECTOR(px)[i] += VECTOR(px)[i-1];
sumpx += VECTOR(px)[i];
}
VECTOR(cpx)[0] = VECTOR(px)[0];
for (i=1; i<il; i++) {
VECTOR(cpx)[i] = VECTOR(cpx)[i-1] + VECTOR(px)[i];
}
for (i=0; i<noiso; i++) {
int ilen=VECTOR(isolen)[i];
int r1= ilen >= fl ? ilen - fl + 1 : 0;
int r2= ilen >= fs ? (ilen >= fl ? fl - fs : ilen - fs + 1) : 0;
/* int r3= fs - 1; */
double sp=0.0;
if (r1 > 0) { sp += r1; }
if (r2 > 0) { sp += VECTOR(cpx)[r2-1]; }
VECTOR(*mysampleprob)[i] = sp * VECTOR(*expression)[i];
if (VECTOR(*mysampleprob)[i] != 0) { goodiso += 1; }
sumpsi += VECTOR(*mysampleprob)[i];
}
if (goodiso == 0) {
SPLICING_ERROR("No isoform is possible", SPLICING_FAILURE);
}
for (i=1; i<noiso; i++) {
VECTOR(*mysampleprob)[i] += VECTOR(*mysampleprob)[i-1];
}
SPLICING_CHECK(splicing_vector_int_resize(isoform, noreads*2));
for (i=0; i<2*noreads; i+=2) {
int w;
double rand;
if (noiso==1) {
w=0;
} else if (noiso==2) {
rand = RNG_UNIF01() * sumpsi;
w = (rand < VECTOR(*mysampleprob)[0]) ? 0 : 1;
} else {
rand = RNG_UNIF01() * sumpsi;
for (w=0; rand > VECTOR(*mysampleprob)[w]; w++) ;
}
VECTOR(*isoform)[i]=VECTOR(*isoform)[i+1]=w;
}
if (!sampleprob) {
splicing_vector_destroy(mysampleprob);
SPLICING_FINALLY_CLEAN(1);
} else {
for (i=noiso-1; i>0; i--) {
VECTOR(*mysampleprob)[i] -= VECTOR(*mysampleprob)[i-1];
}
}
/* We have the isoforms, now get the read positions. */
SPLICING_CHECK(splicing_vector_int_resize(position, noreads*2));
SPLICING_CHECK(splicing_vector_int_init(&exstart, 0));
SPLICING_FINALLY(splicing_vector_int_destroy, &exstart);
SPLICING_CHECK(splicing_vector_int_init(&exend, 0));
SPLICING_FINALLY(splicing_vector_int_destroy, &exend);
SPLICING_CHECK(splicing_vector_int_init(&exidx, 0));
SPLICING_FINALLY(splicing_vector_int_destroy, &exidx);
SPLICING_CHECK(splicing_gff_exon_start_end(gff, &exstart, &exend, &exidx,
gene));
/* Positions in isoform coordinates first.
These are sampled based on the fragment length distribution. */
for (i=0, j=0; i<noreads; i++) {
int iso=VECTOR(*isoform)[2*i];
int ilen=VECTOR(isolen)[iso];
int r1= ilen >= fl ? ilen - fl + 1 : 0;
int r2= ilen >= fs ? (ilen >= fl ? fl - fs : ilen - fs + 1) : 0;
/* int r3= fs - 1; */
int pos, fragment;
double sp=0.0;
if (r1 > 0) { sp += r1; }
if (r2 > 0) { sp += VECTOR(cpx)[r2-1]; }
double rand=RNG_UNIF(0, sp);
if (rand < r1) {
pos = ceil(rand);
} else {
int w;
rand -= r1;
for (w=0; VECTOR(cpx)[w] < rand; w++) ;
pos = r1 + r2 - w;
}
if (pos <= r1) {
rand=RNG_UNIF(0, 1.0);
} else {
rand=RNG_UNIF(0, VECTOR(px)[r1+r2-pos]);
}
for (fragment=0; VECTOR(px)[fragment] < rand; fragment++) ;
fragment += fragmentStart;
VECTOR(*position)[j++] = pos;
VECTOR(*position)[j++] = pos+fragment-readLength;
}
/* Translate positions to genomic coordinates */
/* TODO: some of this is already calculated */
SPLICING_CHECK(splicing_iso_to_genomic(gff, gene, isoform, /*converter=*/ 0,
position));
/* CIGAR strings */
splicing_strvector_clear(cigar);
SPLICING_CHECK(splicing_strvector_reserve(cigar, 2*noreads));
for (j=0; j<2*noreads; j++) {
char tmp[1000], *tmp2=tmp;
int iso=VECTOR(*isoform)[j];
size_t rs=VECTOR(*position)[j];
int ex=0;
int rl=readLength;
for (ex=VECTOR(exidx)[iso]; VECTOR(exend)[ex] < rs; ex++) ;
while (rs + rl - 1 > VECTOR(exend)[ex]) {
tmp2 += snprintf(tmp2, sizeof(tmp)/sizeof(char)-(tmp2-tmp)-1, "%iM%iN",
(int) (VECTOR(exend)[ex]-rs+1),
(int) (VECTOR(exstart)[ex+1]-VECTOR(exend)[ex]-1));
if (tmp2 >= tmp + sizeof(tmp)/sizeof(char)) {
SPLICING_ERROR("CIGAR string too long", SPLICING_EINVAL);
}
rl -= (VECTOR(exend)[ex] - rs + 1);
rs = VECTOR(exstart)[ex+1];
ex++;
}
tmp2 += snprintf(tmp2, sizeof(tmp)/sizeof(char)-(tmp2-tmp)-1, "%iM", rl);
if (tmp2 >= tmp + sizeof(tmp)/sizeof(char)) {
SPLICING_ERROR("CIGAR string too long", SPLICING_EINVAL);
}
SPLICING_CHECK(splicing_strvector_append(cigar, tmp));
}
splicing_vector_int_destroy(&exidx);
splicing_vector_int_destroy(&exend);
splicing_vector_int_destroy(&exstart);
splicing_vector_destroy(&cpx);
splicing_vector_destroy(&px);
splicing_vector_int_destroy(&isolen);
SPLICING_FINALLY_CLEAN(6);
if (!fragmentProb) {
splicing_vector_destroy(myfragmentProb);
SPLICING_FINALLY_CLEAN(1);
}
return 0;
}