static int compute_statistics(GtFeatureNode *fn, void *data, GtError *err)
{
GtStatVisitor *sv;
GT_UNUSED int rval;
gt_error_check(err);
gt_assert(data);
sv = (GtStatVisitor*) data;
if (gt_feature_node_is_multi(fn) &&
gt_feature_node_get_multi_representative(fn) == fn) {
sv->number_of_multi_features++;
}
if (sv->used_sources)
compute_source_statistics(fn, sv->used_sources);
compute_type_statistics(fn, sv);
if (sv->exon_number_distribution || sv->cds_length_distribution) {
sv->exon_number_for_distri = 0;
sv->cds_length_for_distri = 0;
rval = gt_feature_node_traverse_direct_children(fn, sv,
add_exon_or_cds_number,
err);
gt_assert(!rval); /* add_exon_or_cds_number() is sane */
if (sv->exon_number_distribution && sv->exon_number_for_distri) {
gt_disc_distri_add(sv->exon_number_distribution,
sv->exon_number_for_distri);
}
if (sv->cds_length_distribution && sv->cds_length_for_distri) {
gt_disc_distri_add(sv->cds_length_distribution,
sv->cds_length_for_distri);
}
}
return 0;
}
void gth_stat_add_to_sa_alignment_score_distri(GthStat *stat,
unsigned long data)
{
gt_assert(stat);
if (stat->sa_stats)
gt_disc_distri_add(stat->sa_alignment_score_distribution, data);
}
static void gthinitfragments(GtFragment *fragments,
GtUword *num_of_fragments,
GthMatch *storematchtab,
GtUword numofmatches,
GtUword rare,
double fragweightfactor)
{
GthMatch *mptr;
GtFragment *fragmentptr;
GtWord tmp, largestdim1 = 0, largestdim2 = 0;
GtDiscDistri *startpointdistri = NULL;
/* init number of fragments */
*num_of_fragments = 0;
if (rare)
startpointdistri = gt_disc_distri_new();
for (mptr = storematchtab; mptr < storematchtab + numofmatches; mptr++) {
/* first dimension */
tmp = mptr->Storepositionreference + mptr->Storelengthreference - 1;
if (largestdim1 < tmp)
largestdim1 = tmp;
/* second dimension */
tmp = mptr->Storepositiongenomic + mptr->Storelengthgenomic - 1;
if (largestdim2 < tmp)
largestdim2 = tmp;
}
for (mptr = storematchtab, fragmentptr = fragments;
mptr < storematchtab + numofmatches;
mptr++) {
if (rare)
gt_disc_distri_add(startpointdistri, mptr->Storepositionreference);
if ((!rare ||
gt_disc_distri_get(startpointdistri, mptr->Storepositionreference)
<= rare) &&
(mptr == storematchtab || /* is the first match */
!gth_matches_are_equal(mptr, mptr-1))) { /* or is different from last
one */
fragmentptr->weight = (GtWord) (fragweightfactor *
(double) abs(mptr->Storescore));
fragmentptr->startpos1 = mptr->Storepositionreference;
fragmentptr->endpos1 = mptr->Storepositionreference
+ mptr->Storelengthreference - 1;
fragmentptr->startpos2 = mptr->Storepositiongenomic;
fragmentptr->endpos2 = mptr->Storepositiongenomic
+ mptr->Storelengthgenomic - 1;
fragmentptr++;
(*num_of_fragments)++;
}
}
gt_disc_distri_delete(startpointdistri);
gt_assert(*num_of_fragments <= numofmatches);
}
static void compute_type_statistics(GtFeatureNode *fn, GtStatVisitor *sv)
{
GtRange range;
gt_assert(fn && sv);
if (gt_feature_node_has_type(fn, gt_ft_gene)) {
sv->number_of_genes++;
if (gt_feature_node_has_CDS(fn))
sv->number_of_protein_coding_genes++;
if (sv->gene_length_distribution) {
range = gt_genome_node_get_range((GtGenomeNode*) fn);
gt_disc_distri_add(sv->gene_length_distribution, gt_range_length(&range));
}
if (sv->gene_score_distribution) {
gt_disc_distri_add(sv->gene_score_distribution,
gt_feature_node_get_score(fn) * 100.0);
}
}
else if (gt_feature_node_has_type(fn, gt_ft_mRNA)) {
sv->number_of_mRNAs++;
if (gt_feature_node_has_CDS(fn))
sv->number_of_protein_coding_mRNAs++;
}
else if (gt_feature_node_has_type(fn, gt_ft_exon)) {
sv->number_of_exons++;
if (sv->exon_length_distribution) {
range = gt_genome_node_get_range((GtGenomeNode*) fn);
gt_disc_distri_add(sv->exon_length_distribution,
gt_range_length(&range));
}
}
else if (gt_feature_node_has_type(fn, gt_ft_CDS)) {
sv->number_of_CDSs++;
}
else if (gt_feature_node_has_type(fn, gt_ft_intron)) {
if (sv->intron_length_distribution) {
range = gt_genome_node_get_range((GtGenomeNode*) fn);
gt_disc_distri_add(sv->intron_length_distribution,
gt_range_length(&range));
}
}
else if (gt_feature_node_has_type(fn, gt_ft_LTR_retrotransposon)) {
sv->number_of_LTR_retrotransposons++;
}
}
GtDiscDistri *gt_condenseq_link_length_dist(const GtCondenseq *condenseq)
{
GtUword idx;
GtDiscDistri *res = gt_disc_distri_new();
for (idx = 0; idx < condenseq->ldb_nelems; idx++) {
gt_disc_distri_add(res, condenseq->links[idx].len);
}
return res;
}
void gt_bioseq_show_seqlengthdistri(GtBioseq *bs, GtFile *outfp)
{
GtDiscDistri *d;
GtUword i;
gt_assert(bs);
d = gt_disc_distri_new();
for (i = 0; i < gt_bioseq_number_of_sequences(bs); i++)
gt_disc_distri_add(d, gt_bioseq_get_sequence_length(bs, i));
gt_file_xprintf(outfp, "sequence length distribution:\n");
gt_disc_distri_show(d, outfp);
gt_disc_distri_delete(d);
}
static void addSAtoexondistribution(GtDiscDistri *exondistribution,
GthSA *sa)
{
Exoninfo *exoninfo;
unsigned long i;
/* add values to exondistribution */
for (i = 0; i < gth_sa_num_of_exons(sa); i++) {
exoninfo = gth_sa_get_exon(sa, i);
gt_disc_distri_add(exondistribution, exoninfo->rightgenomicexonborder -
exoninfo->leftgenomicexonborder + 1);
}
}
static void addSAtointrondistribution(GtDiscDistri *introndistribution,
GthSA *sa)
{
unsigned long i;
/* add values to introndistribution */
for (i = 0; i < gth_sa_num_of_introns(sa); i++) {
gt_disc_distri_add(introndistribution,
gth_sa_get_exon(sa, i+1)
->leftgenomicexonborder -
gth_sa_get_exon(sa, i)
->rightgenomicexonborder - 1);
}
}
GtDiscDistri *gt_condenseq_link_comp_dist(const GtCondenseq *condenseq)
{
GtUword idx;
GtDiscDistri *res = gt_disc_distri_new();
for (idx = 0; idx < condenseq->ldb_nelems; idx++) {
GtEditscript *es = condenseq->links[idx].editscript;
GtUword vlen;
size_t size;
vlen = gt_editscript_get_target_len(es);
size = gt_editscript_size(es);
gt_disc_distri_add(res, (GtUword) ((double) size/(double) vlen * 100));
}
return res;
}
int gt_disc_distri_unit_test(GtError *err)
{
GtDiscDistri *d;
int had_err = 0;
gt_error_check(err);
d = gt_disc_distri_new();
ensure(had_err, gt_disc_distri_get(d, 0) == 0);
ensure(had_err, gt_disc_distri_get(d, 100) == 0);
if (!had_err) {
gt_disc_distri_add(d, 0);
gt_disc_distri_add_multi(d, 100, 256);
}
ensure(had_err, gt_disc_distri_get(d, 0) == 1);
ensure(had_err, gt_disc_distri_get(d, 100) == 256);
gt_disc_distri_delete(d);
return had_err;
}
static void outlcpvalues(Lcpsubtab *lcpsubtab,
GtUword width,
GtUword posoffset)
{
GtUword idx, lcpvalue;
Largelcpvalue *largelcpvalueptr;
gt_assert(lcpsubtab != NULL && lcpsubtab->lcp2file != NULL);
lcpsubtab->lcp2file->largelcpvalues.nextfreeLargelcpvalue = 0;
if (lcpsubtab->tableoflcpvalues.numoflargelcpvalues > 0 &&
lcpsubtab->tableoflcpvalues.numoflargelcpvalues >=
lcpsubtab->lcp2file->largelcpvalues.allocatedLargelcpvalue)
{
lcpsubtab->lcp2file->largelcpvalues.spaceLargelcpvalue
= gt_realloc(lcpsubtab->lcp2file->largelcpvalues.spaceLargelcpvalue,
sizeof (*lcpsubtab->lcp2file->largelcpvalues.
spaceLargelcpvalue) *
lcpsubtab->tableoflcpvalues.numoflargelcpvalues);
lcpsubtab->lcp2file->largelcpvalues.allocatedLargelcpvalue
= lcpsubtab->tableoflcpvalues.numoflargelcpvalues;
}
for (idx=0; idx<width; idx++)
{
lcpvalue = gt_lcptab_getvalue(&lcpsubtab->tableoflcpvalues,0,idx);
if (lcpsubtab->lcp2file->maxbranchdepth < lcpvalue)
{
lcpsubtab->lcp2file->maxbranchdepth = lcpvalue;
}
if (lcpvalue < (GtUword) LCPOVERFLOW)
{
lcpsubtab->lcp2file->smalllcpvalues[idx] = (uint8_t) lcpvalue;
} else
{
gt_assert(lcpsubtab->lcp2file->largelcpvalues.nextfreeLargelcpvalue
< lcpsubtab->lcp2file->largelcpvalues.
allocatedLargelcpvalue);
largelcpvalueptr
= lcpsubtab->lcp2file->largelcpvalues.spaceLargelcpvalue +
lcpsubtab->lcp2file->largelcpvalues.nextfreeLargelcpvalue++;
largelcpvalueptr->position = posoffset + idx;
largelcpvalueptr->value = lcpvalue;
lcpsubtab->lcp2file->smalllcpvalues[idx] = LCPOVERFLOW;
}
lcpsubtab->lcptabsum += (double) lcpvalue;
if (lcpsubtab->distlcpvalues != NULL)
{
gt_disc_distri_add(lcpsubtab->distlcpvalues, lcpvalue);
}
}
outsmalllcpvalues(lcpsubtab->lcp2file,width);
if (lcpsubtab->lcp2file->largelcpvalues.nextfreeLargelcpvalue > 0)
{
lcpsubtab->lcp2file->totalnumoflargelcpvalues
+= lcpsubtab->lcp2file->largelcpvalues.nextfreeLargelcpvalue;
gt_assert(lcpsubtab->lcp2file->outfpllvtab != NULL);
gt_xfwrite(lcpsubtab->lcp2file->largelcpvalues.spaceLargelcpvalue,
sizeof (*lcpsubtab->lcp2file->largelcpvalues.
spaceLargelcpvalue),
(size_t) lcpsubtab->lcp2file->largelcpvalues.
nextfreeLargelcpvalue,
lcpsubtab->lcp2file->outfpllvtab);
}
}
void gth_stat_add_to_refseqcovdistri(GthStat *stat, unsigned long data)
{
gt_assert(stat);
if (stat->refseqcovdistri)
gt_disc_distri_add(stat->refseqcoveragedistribution, data);
}
void gth_stat_add_to_matchnumdistri(GthStat *stat, unsigned long data)
{
gt_assert(stat);
if (stat->matchnumdistri)
gt_disc_distri_add(stat->matchnumdistribution, data);
}
void gth_stat_add_to_sa_coverage_distri(GthStat *stat, GtUword data)
{
gt_assert(stat);
if (stat->sa_stats)
gt_disc_distri_add(stat->sa_coverage_distribution, data);
}
void gth_stat_add_to_matchnumdistri(GthStat *stat, GtUword data)
{
gt_assert(stat);
if (stat->matchnumdistri)
gt_disc_distri_add(stat->matchnumdistribution, data);
}
