int splicing_dgesdd(const splicing_matrix_t *matrix, splicing_vector_t *values) { splicing_matrix_t tmp; int m=splicing_matrix_nrow(matrix); int n=splicing_matrix_ncol(matrix); int lda=m, minmn= m < n ? m : n, maxmn = m < n ? n : m; int lwork=-1; int info=0; splicing_vector_t work; splicing_vector_int_t iwork; char jobz='N'; int dummy=1; double dummy2; SPLICING_CHECK(splicing_matrix_copy(&tmp, matrix)); SPLICING_FINALLY(splicing_matrix_destroy, &tmp); SPLICING_CHECK(splicing_vector_init(&work, 1)); SPLICING_FINALLY(splicing_vector_destroy, &work); SPLICING_CHECK(splicing_vector_int_init(&iwork, 8*minmn)); SPLICING_FINALLY(splicing_vector_int_destroy, &iwork); SPLICING_CHECK(splicing_vector_resize(values, minmn)); /* Get the optiomal lwork first*/ splicingdgesdd_(&jobz, &m, &n, &MATRIX(tmp,0,0), &lda, VECTOR(*values), /*U=*/ &dummy2, /*LDU=*/ &dummy, /*VT=*/ &dummy2, /*LDVT=*/ &dummy, VECTOR(work), &lwork, VECTOR(iwork), &info); lwork = VECTOR(work)[0]; SPLICING_CHECK(splicing_vector_resize(&work, lwork)); /* Now do the SVD */ splicingdgesdd_(&jobz, &m, &n, &MATRIX(tmp,0,0), &lda, VECTOR(*values), /*U=*/ &dummy2, /*LDU=*/ &dummy, /*VT=*/ &dummy2, /*LDVT=*/ &dummy, VECTOR(work), &lwork, VECTOR(iwork), &info); if (info != 0) { SPLICING_ERROR("Cannot calculate SVD", SPLICING_ELAPACK); } splicing_vector_destroy(&work); splicing_vector_int_destroy(&iwork); splicing_matrix_destroy(&tmp); SPLICING_FINALLY_CLEAN(3); return 0; }
int splicing_io_get_string(FILE *input, char *buffer, size_t maxlen, size_t *len, char delim, int newline) { int c; *len = 0; while (1) { c=fgetc(input); if (c==EOF) { SPLICING_ERROR("Invalid GFF file", SPLICING_PARSEERROR); } else if (*len == maxlen) { SPLICING_ERROR("Invalid GFF file", SPLICING_PARSEERROR); } else if (c==delim) { *buffer='\0'; buffer++; return 0; } else if (newline && (c=='\n' || c=='\r')) { *buffer='\0'; buffer++; return 0; } else if (!newline && (c=='\n' || c=='\r')) { SPLICING_ERROR("Invalid GFF file", SPLICING_PARSEERROR); } else { *buffer=(char) c; buffer++; *len += 1; } } return 1; }
int splicing_io_get_real_na(FILE *input, double *real, char delim, char nachar) { char buffer[30]; char *bufend; size_t len; double na=SPLICING_NA_REAL; int eof = splicing_io_get_string(input, buffer, sizeof(buffer)/sizeof(char), &len, delim, /*newline=*/ 0); if (eof) { SPLICING_ERROR("Invalid GFF file", SPLICING_PARSEERROR); } if (len > 0 && buffer[0]==nachar) { *real=na; return 0; } *real = strtod(buffer, &bufend); if (*bufend != '\0') { SPLICING_ERROR("Invalid GFF file", SPLICING_PARSEERROR); } return 0; }
int splicing_io_get_integer_na(FILE *input, int *integer, char delim, char nachar) { char buffer[30]; char *bufend; size_t len; int na=SPLICING_NA_INTEGER; int eof = splicing_io_get_string(input, buffer, sizeof(buffer)/sizeof(char), &len, delim, /*newline=*/ 0); if (eof) { SPLICING_ERROR("Invalid GFF file", SPLICING_PARSEERROR); } if (len > 0 && buffer[0]==nachar) { *integer=na; return 0; } *integer = (int) strtol(buffer, &bufend, /*base=*/ 10); if (*bufend != '\0') { SPLICING_ERROR("Invalid GFF file", SPLICING_PARSEERROR); } return 0; }
int splicing_io_parse_attributes(char *attr, char **ID, char**parent) { *ID=SPLICING_STRVECTOR_ZERO; *parent=SPLICING_STRVECTOR_ZERO; char *kw, *vl; while (*attr != '\0') { /* Skip white space */ while (*attr != '\0' && isspace(*attr)) { attr++; } /* Keyword */ kw=attr; while (*attr != '\0' && *attr != '=') { attr++; } if (*attr == '\0') { SPLICING_ERROR("Invalid GFF file", SPLICING_PARSEERROR); } *attr='\0'; attr++; /* Value */ vl=attr; while (*attr != '\0' && *attr != ';') { attr++; } if (*attr == ';') { *attr='\0'; attr++; } if (!strcmp("ID", kw)) { *ID=vl; } else if (!strcmp("Parent", kw)) { *parent=vl; } } return 0; }
int splicing_gff_noexons_one(const splicing_gff_t *gff, size_t gene, splicing_vector_int_t *noexons) { size_t nogenes, idx1, idx2, noiso, pos, il; SPLICING_CHECK(splicing_gff_nogenes(gff, &nogenes)); if (gene >= nogenes) { SPLICING_ERROR("Invalid gene id", SPLICING_EINVAL); } idx1=VECTOR(gff->genes)[gene]; idx2= gene+1 == nogenes ? gff->n : VECTOR(gff->genes)[gene+1]; for (noiso=0; idx1 < idx2; idx1++) { if (VECTOR(gff->type)[idx1] == SPLICING_TYPE_MRNA) { noiso += 1; } } SPLICING_CHECK(splicing_vector_int_resize(noexons, noiso)); idx1=VECTOR(gff->genes)[gene]; idx2= gene+1 == nogenes ? gff->n : VECTOR(gff->genes)[gene+1]; for (; idx1 < idx2 && VECTOR(gff->type)[idx1] != SPLICING_TYPE_MRNA; idx1++) ; idx1++; for (pos=0, il=0; idx1 < idx2; idx1++) { if (VECTOR(gff->type)[idx1] == SPLICING_TYPE_MRNA) { VECTOR(*noexons)[pos++]=il; il=0; } else if (VECTOR(gff->type)[idx1] == SPLICING_TYPE_EXON) { il++; } } VECTOR(*noexons)[pos++]=il; 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; }
/* TODO: do not ignore size */ int splicing_gff_init(splicing_gff_t *gff, size_t size) { if (size < 0) { SPLICING_ERROR("Cannot create GFF, `size' must be non-negative", SPLICING_EINVAL); } SPLICING_CHECK(splicing_strvector_init(&gff->seqids, 0)); SPLICING_FINALLY(splicing_strvector_destroy, &gff->seqids); SPLICING_CHECK(splicing_strvector_init(&gff->sources, 0)); SPLICING_FINALLY(splicing_strvector_destroy, &gff->sources); SPLICING_CHECK(splicing_strvector_init(&gff->ID, 0)); SPLICING_FINALLY(splicing_strvector_destroy, &gff->ID); SPLICING_CHECK(splicing_vector_int_init(&gff->genes, 0)); SPLICING_FINALLY(splicing_vector_int_destroy, &gff->genes); SPLICING_CHECK(splicing_vector_int_init(&gff->transcripts, 0)); SPLICING_FINALLY(splicing_vector_int_destroy, &gff->transcripts); SPLICING_CHECK(splicing_vector_int_init(&gff->seqid, 0)); SPLICING_FINALLY(splicing_vector_int_destroy, &gff->seqid); SPLICING_CHECK(splicing_vector_int_init(&gff->source, 0)); SPLICING_FINALLY(splicing_vector_int_destroy, &gff->source); SPLICING_CHECK(splicing_vector_int_init(&gff->strand, 0)); SPLICING_FINALLY(splicing_vector_int_destroy, &gff->strand); SPLICING_CHECK(splicing_vector_int_init(&gff->type, 0)); SPLICING_FINALLY(splicing_vector_int_destroy, &gff->type); SPLICING_CHECK(splicing_vector_int_init(&gff->start, 0)); SPLICING_FINALLY(splicing_vector_int_destroy, &gff->start); SPLICING_CHECK(splicing_vector_int_init(&gff->end, 0)); SPLICING_FINALLY(splicing_vector_int_destroy, &gff->end); SPLICING_CHECK(splicing_vector_init(&gff->score, 0)); SPLICING_FINALLY(splicing_vector_destroy, &gff->score); SPLICING_CHECK(splicing_vector_int_init(&gff->phase, 0)); SPLICING_FINALLY(splicing_vector_int_destroy, &gff->phase); SPLICING_CHECK(splicing_vector_int_init(&gff->parent, 0)); SPLICING_FINALLY(splicing_vector_int_destroy, &gff->parent); gff->n=0; gff->nogenes=0; gff->notranscripts=0; gff->last_gene_id = gff->last_mrna_id = SPLICING_STRVECTOR_ZERO; gff->last_gene_no = gff->last_mrna_no = -1; gff->last_seqid = gff->last_source = SPLICING_STRVECTOR_ZERO; SPLICING_FINALLY_CLEAN(14); return 0; }
int splicing_gff_gene_start_end_one(const splicing_gff_t *gff, size_t gene, size_t *start, size_t *end) { size_t nogenes=splicing_vector_int_size(&gff->genes); size_t idx; if (gene < 0 || gene >= nogenes) { SPLICING_ERROR("Invalid gene id", SPLICING_EINVAL); } idx=VECTOR(gff->genes)[gene]; *start=VECTOR(gff->start)[idx]; *end=VECTOR(gff->end)[idx]; return 0; }
int splicing_gff_constitutive_exons(const splicing_gff_t *gff, splicing_exonset_t *exons, int min_length, splicing_constitutive_mode_t mode) { switch (mode) { case SPLICING_CONSTITUTIVE_ALL: return splicing_i_gff_constitutive_exons_all(gff, exons, min_length); break; case SPLICING_CONSTITUTIVE_FULL: return splicing_i_gff_constitutive_exons_full(gff, exons, min_length); break; default: SPLICING_ERROR("Unknown `mode' argument for constitutive exon finding", SPLICING_EINVAL); } }
int splicing_i_gff_reindex_cmp(void *data, const void *a, const void *b) { splicing_gff_t *gff=(splicing_gff_t *) data; int aa=*(int*)a, bb=*(int*)b; int parent_a=VECTOR(gff->parent)[aa]; int parent_b=VECTOR(gff->parent)[bb]; int gparent_a= parent_a == -1 ? -1 : VECTOR(gff->parent)[parent_a]; int gparent_b= parent_b == -1 ? -1 : VECTOR(gff->parent)[parent_b]; const char *a_gene_id, *b_gene_id, *a_mrna_id, *b_mrna_id; int c1, c2; /* If gene ids differ */ a_gene_id = gparent_a != -1 ? STR(gparent_a) : (parent_a != -1 ? STR(parent_a) : STR(aa)); b_gene_id = gparent_b != -1 ? STR(gparent_b) : (parent_b != -1 ? STR(parent_b) : STR(bb)); c1=strcmp(a_gene_id, b_gene_id); if (c1 != 0) { return c1; } /* Or if mRNA ids differ */ a_mrna_id = gparent_a != -1 ? STR(parent_a) : STR(aa); b_mrna_id = gparent_b != -1 ? STR(parent_b) : STR(bb); c2=strcmp(a_mrna_id, b_mrna_id); if (c2 != 0) { return c2; } /* Otherwise gene first, then mRNA, then the rest according to start position */ if (parent_a == -1 && parent_b != -1) { return -1; } else if (parent_a != -1 && parent_b == -1) { return 1; } else if (gparent_a == -1 && gparent_b != -1) { return -1; } else if (gparent_a != -1 && gparent_b == -1) { return 1; } else if (gparent_a != -1 && gparent_b != -1) { int sa=VECTOR(gff->start)[aa]; int sb=VECTOR(gff->start)[bb]; if (sa < sb) { return -1; } else if (sa > sb) { return 1; } return 0; } else { SPLICING_ERROR("Invalid GFF file, cannot order records", SPLICING_EINVAL); } return 0; }
int splicing_iso_to_genomic_all(const splicing_gff_t *gff, size_t gene, int position, const splicing_gff_converter_t *converter, splicing_vector_int_t *result) { size_t i; splicing_gff_converter_t vconverter, *myconverter = (splicing_gff_converter_t*) converter; if (position < 1) { SPLICING_ERROR("Invalid isoform coordinate, must the larger than zero", SPLICING_EINVAL); } if (!converter) { myconverter=&vconverter; SPLICING_CHECK(splicing_gff_converter_init(gff, gene, myconverter)); SPLICING_FINALLY(splicing_gff_converter_destroy, myconverter); } SPLICING_CHECK(splicing_vector_int_resize(result, myconverter->noiso)); /* TODO: find impossible positions */ for (i=0; i<myconverter->noiso; i++) { int ex; for (ex=VECTOR(myconverter->exidx)[i]; ex < VECTOR(myconverter->exidx)[i+1] && VECTOR(myconverter->exlim)[ex] <= position; ex++) ; if (ex < VECTOR(myconverter->exidx)[i+1]) { VECTOR(*result)[i] = position + VECTOR(myconverter->shift)[ex]; } else { VECTOR(*result)[i] = -1; } } if (!converter) { splicing_gff_converter_destroy(myconverter); SPLICING_FINALLY_CLEAN(1); } return 0; }
int splicing_i_gff_noiso_one(const splicing_gff_t *gff, size_t gene, size_t *noiso, splicing_vector_int_t *isolen) { size_t nogenes, idx1, idx2; SPLICING_CHECK(splicing_gff_nogenes(gff, &nogenes)); if (gene < 0 || gene >= nogenes) { SPLICING_ERROR("Invalid gene id", SPLICING_EINVAL); } idx1=VECTOR(gff->genes)[gene]; idx2= gene+1 == nogenes ? gff->n : VECTOR(gff->genes)[gene+1]; *noiso = 0; for ( ; idx1 < idx2; idx1++) { if (VECTOR(gff->type)[idx1] == SPLICING_TYPE_MRNA) { *noiso += 1; } } if (isolen) { size_t il=0, pos=0; SPLICING_CHECK(splicing_vector_int_resize(isolen, *noiso)); idx1=VECTOR(gff->genes)[gene]; idx2= gene+1 == nogenes ? gff->n : VECTOR(gff->genes)[gene+1]; for (; idx1 < idx2 && VECTOR(gff->type)[idx1] != SPLICING_TYPE_MRNA; idx1++) ; idx1++; for (; idx1 < idx2; idx1++) { if (VECTOR(gff->type)[idx1] == SPLICING_TYPE_MRNA) { VECTOR(*isolen)[pos++]=il; il = 0; } else if (VECTOR(gff->type)[idx1] == SPLICING_TYPE_EXON) { il += VECTOR(gff->end)[idx1] - VECTOR(gff->start)[idx1] + 1; } } VECTOR(*isolen)[pos++]=il; } return 0; }
int splicing_gff_fprint_gene(const splicing_gff_t *gff, FILE *outfile, int gene) { size_t nogenes, noiso; int i, j; splicing_vector_int_t start, end, idx; SPLICING_CHECK(splicing_gff_nogenes(gff, &nogenes)); if (gene < 0 || gene >= nogenes) { SPLICING_ERROR("Invalid gene ID", SPLICING_EINVAL); } SPLICING_CHECK(splicing_vector_int_init(&start, 0)); SPLICING_FINALLY(splicing_vector_int_destroy, &start); SPLICING_CHECK(splicing_vector_int_init(&end, 0)); SPLICING_FINALLY(splicing_vector_int_destroy, &end); SPLICING_CHECK(splicing_vector_int_init(&idx, 0)); SPLICING_FINALLY(splicing_vector_int_destroy, &idx); SPLICING_CHECK(splicing_gff_exon_start_end(gff, &start, &end, &idx, gene)); noiso = splicing_vector_int_size(&idx)-1; fprintf(outfile, "===\nGene with %i isoforms:\n", (int) noiso); for (i=0; i<noiso; i++) { fprintf(outfile, " Isoform %i:\n", i); for (j=VECTOR(idx)[i]; j<VECTOR(idx)[i+1]; j++) { fprintf(outfile, " %i-%i\n", VECTOR(start)[j], VECTOR(end)[j]); } } splicing_vector_int_destroy(&idx); splicing_vector_int_destroy(&end); splicing_vector_int_destroy(&start); SPLICING_FINALLY_CLEAN(3); return 0; }
int splicing_miso_trinity(const splicing_matrix_t *match_matrix, const splicing_vector_int_t *isolen, int readLength, int noIterations, int noBurnIn, int noLag, const splicing_vector_t *hyperp, 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) { double acceptP, cJS, pJS, sigma; int noiso = splicing_matrix_nrow(match_matrix); int noReads = splicing_matrix_ncol(match_matrix); splicing_vector_int_t *myass=assignment, vass; 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_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); } 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); SPLICING_CHECK(splicing_vector_int_init(&match_order, noReads)); SPLICING_FINALLY(splicing_vector_int_destroy, &match_order); SPLICING_CHECK(splicing_order_matches(match_matrix, &match_order)); if (class_templates && class_counts) { SPLICING_CHECK(splicing_i_miso_classes(match_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); for (i=0; i<noiso; i++) { int l=VECTOR(*isolen)[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(match_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(match_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_vector_destroy(&valphaNew); splicing_vector_destroy(&valpha); splicing_vector_destroy(&vpsiNew); splicing_vector_destroy(&vpsi); SPLICING_FINALLY_CLEAN(7); if (!assignment) { splicing_vector_int_destroy(myass); 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; }
int splicing_gff_read(FILE *input, splicing_gff_t *gff) { int eof=!EOF; char seqid[200]; char source[200]; char type[200]; int start, end, phase; double score; char strand[10]; char attributes[5000]; size_t len; splicing_type_t realtype; splicing_strand_t realstrand=SPLICING_STRAND_UNKNOWN; char *ID, *parent; do { eof = eof || splicing_io_skip_newline_and_comments(input); eof = eof || splicing_io_get_string(input, seqid, sizeof(seqid)/sizeof(char), &len, /*delim=*/ '\t', /*newline=*/ 0); eof = eof || splicing_io_get_string(input, source, sizeof(source)/sizeof(char), &len, /*delim=*/ '\t', /*newline=*/ 0); eof = eof || splicing_io_get_string(input, type, sizeof(type)/sizeof(char), &len, /*delim=*/ '\t', /*newline=*/ 0); eof = eof || splicing_io_get_integer(input, &start, /*delim=*/ '\t'); eof = eof || splicing_io_get_integer(input, &end, /*delim=*/ '\t'); eof = eof || splicing_io_get_real_na(input, &score, /*delim=*/ '\t', /*nachar=*/ '.'); eof = eof || splicing_io_get_string(input, strand, sizeof(strand)/sizeof(char), &len, /*delim=*/ '\t', /*newline=*/ 0); eof = eof || splicing_io_get_integer_na(input, &phase, /*delim=*/ '\t', /*nachar=*/ '.'); eof = eof || splicing_io_get_string(input, attributes, sizeof(attributes)/sizeof(char), &len, /*delim=*/ '\n', /*newline=*/ 1); if (eof) { SPLICING_ERROR("Corrupt GFF file", SPLICING_PARSEERROR); } /* TODO: do not hardcode these names TODO: order them according to their frequency */ if (!strcmp(type, "gene")) { realtype = SPLICING_TYPE_GENE; } else if (!strcmp(type, "mRNA")) { realtype = SPLICING_TYPE_MRNA; } else if (!strcmp(type, "exon")) { realtype = SPLICING_TYPE_EXON; } else if (!strcmp(type, "CDS")) { realtype = SPLICING_TYPE_CDS; } else if (!strcmp(type, "start_codon")) { realtype = SPLICING_TYPE_START_CODON; } else if (!strcmp(type, "stop_codon")) { realtype = SPLICING_TYPE_STOP_CODON; } else { SPLICING_ERROR("Invalid GFF file", SPLICING_PARSEERROR); } if (!strcmp(strand, "+")) { realstrand=SPLICING_STRAND_PLUS; } else if (!strcmp(strand, "-")) { realstrand=SPLICING_STRAND_MINUS; } else if (!strcmp(strand, ".")) { realstrand=SPLICING_STRAND_UNKNOWN; } /* Parsing the attributes field */ SPLICING_CHECK(splicing_io_parse_attributes(attributes, &ID, &parent)); SPLICING_CHECK(splicing_gff_append(gff, seqid, source, realtype, start, end, score, realstrand, phase, ID, parent)); eof = splicing_io_skip_newline_and_comments(input); } while (!eof); return 0; }
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_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; }