int splicing_vector_rank(const splicing_vector_t *v, splicing_vector_t *res, long int nodes) { splicing_vector_t rad; splicing_vector_t ptr; long int edges = splicing_vector_size(v); long int i, c=0; SPLICING_VECTOR_INIT_FINALLY(&rad, nodes); SPLICING_VECTOR_INIT_FINALLY(&ptr, edges); splicing_vector_resize(res, edges); for (i=0; i<edges; i++) { long int elem=VECTOR(*v)[i]; VECTOR(ptr)[i] = VECTOR(rad)[elem]; VECTOR(rad)[elem] = i+1; } for (i=0; i<nodes; i++) { long int p=VECTOR(rad)[i]; while (p != 0) { VECTOR(*res)[p-1]=c++; p=VECTOR(ptr)[p-1]; } } splicing_vector_destroy(&ptr); splicing_vector_destroy(&rad); SPLICING_FINALLY_CLEAN(2); return 0; }
PyObject *pysplicing_from_vector(const splicing_vector_t *v) { int i, n=splicing_vector_size(v); PyObject *o=PyTuple_New(n); for (i=0; i<n; i++) { PyObject *it=PyFloat_FromDouble(VECTOR(*v)[i]); PyTuple_SetItem(o, i, it); } return o; }
int splicing_vector_round(const splicing_vector_t *from, splicing_vector_long_t *to) { long int i, n=splicing_vector_size(from); splicing_vector_long_resize(to, n); for (i=0; i<n; i++) { VECTOR(*to)[i] = round(VECTOR(*from)[i]); } return 0; }
int splicing_vector_order1(const splicing_vector_t* v, splicing_vector_t* res, double nodes) { long int edges=splicing_vector_size(v); splicing_vector_t ptr; splicing_vector_t rad; long int i, j; assert(v!=NULL); assert(v->stor_begin != NULL); SPLICING_VECTOR_INIT_FINALLY(&ptr, nodes+1); SPLICING_VECTOR_INIT_FINALLY(&rad, edges); splicing_vector_resize(res, edges); for (i=0; i<edges; i++) { long int radix=v->stor_begin[i]; if (VECTOR(ptr)[radix]!=0) { VECTOR(rad)[i]=VECTOR(ptr)[radix]; } VECTOR(ptr)[radix]=i+1; } j=0; for (i=0; i<nodes+1; i++) { if (VECTOR(ptr)[i] != 0) { long int next=VECTOR(ptr)[i]-1; res->stor_begin[j++]=next; while (VECTOR(rad)[next] != 0) { next=VECTOR(rad)[next]-1; res->stor_begin[j++]=next; } } } splicing_vector_destroy(&ptr); splicing_vector_destroy(&rad); SPLICING_FINALLY_CLEAN(2); return 0; }
int splicing_gene_complexity(const splicing_gff_t *gff, size_t gene, int readLength, splicing_complexity_t type, splicing_norm_t norm, int paired, const splicing_vector_t *fragmentProb, int fragmentStart, double normalMean, double normalVar, double numDevs, double *complexity) { splicing_matrix_t assignment_matrix; SPLICING_CHECK(splicing_matrix_init(&assignment_matrix, 0, 0)); SPLICING_FINALLY(splicing_matrix_destroy, &assignment_matrix); if (!paired) { SPLICING_CHECK(splicing_assignment_matrix(gff, gene, readLength, &assignment_matrix)); } else { SPLICING_CHECK(splicing_paired_assignment_matrix(gff, gene, readLength, fragmentProb, fragmentStart, normalMean, normalVar, numDevs, &assignment_matrix)); } switch (type) { case SPLICING_COMPLEXITY_RELATIVE: switch (norm) { splicing_vector_t values; int i, n; case SPLICING_NORM_2: SPLICING_CHECK(splicing_vector_init(&values, 0)); SPLICING_FINALLY(splicing_vector_destroy, &values); SPLICING_CHECK(splicing_dgesdd(&assignment_matrix, &values)); n=splicing_vector_size(&values); for (i=n-1; i>=0 && VECTOR(values)[i] < 1e-14; i--) ; *complexity = VECTOR(values)[0] / VECTOR(values)[i]; splicing_vector_destroy(&values); SPLICING_FINALLY_CLEAN(1); break; case SPLICING_NORM_1: SPLICING_ERROR("One norm not implemented", SPLICING_UNIMPLEMENTED); break; case SPLICING_NORM_INFINITY: SPLICING_ERROR("Infinity norm not implemented", SPLICING_UNIMPLEMENTED); break; } break; case SPLICING_COMPLEXITY_ABSOLUTE: SPLICING_ERROR("Absolute complexity not implemented", SPLICING_UNIMPLEMENTED); break; } splicing_matrix_destroy(&assignment_matrix); SPLICING_FINALLY_CLEAN(1); 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; }