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);
}
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);
}
void gt_Outlcpinfo_delete(GtOutlcpinfo *outlcpinfo)
{
if (outlcpinfo == NULL)
{
return;
}
gt_turningwheel_delete(outlcpinfo->turnwheel);
if (outlcpinfo->lcpsubtab.lcp2file != NULL)
{
if (!outlcpinfo->swallow_tail_lcpvalues &&
outlcpinfo->lcpsubtab.lcp2file->countoutputlcpvalues <
outlcpinfo->numsuffixes2output)
{
outlcpinfo->lcpsubtab.lcp2file->countoutputlcpvalues
+= outmany0lcpvalues(outlcpinfo->numsuffixes2output -
outlcpinfo->lcpsubtab.lcp2file
->countoutputlcpvalues,
outlcpinfo->lcpsubtab.lcp2file->outfplcptab);
}
gt_assert(outlcpinfo->swallow_tail_lcpvalues ||
outlcpinfo->lcpsubtab.lcp2file->countoutputlcpvalues ==
outlcpinfo->numsuffixes2output);
GT_FREEARRAY(&outlcpinfo->lcpsubtab.lcp2file->largelcpvalues,
Largelcpvalue);
gt_fa_fclose(outlcpinfo->lcpsubtab.lcp2file->outfplcptab);
gt_fa_fclose(outlcpinfo->lcpsubtab.lcp2file->outfpllvtab);
gt_free(outlcpinfo->lcpsubtab.lcp2file->reservoir);
outlcpinfo->lcpsubtab.lcp2file->smalllcpvalues = NULL;
outlcpinfo->lcpsubtab.lcp2file->reservoir = NULL;
outlcpinfo->lcpsubtab.lcp2file->sizereservoir = 0;
gt_free(outlcpinfo->lcpsubtab.lcp2file);
} else
{
gt_free(outlcpinfo->lcpsubtab.tableoflcpvalues.bucketoflcpvalues);
#ifndef NDEBUG
gt_free(outlcpinfo->lcpsubtab.tableoflcpvalues.isset);
#endif
}
gt_free(outlcpinfo->lcpsubtab.lcpprocess);
outlcpinfo->lcpsubtab.tableoflcpvalues.bucketoflcpvalues = NULL;
#ifndef NDEBUG
outlcpinfo->lcpsubtab.tableoflcpvalues.isset = NULL;
#endif
outlcpinfo->lcpsubtab.tableoflcpvalues.numofentries = 0;
if (outlcpinfo->lcpsubtab.distlcpvalues != NULL)
{
gt_disc_distri_show(outlcpinfo->lcpsubtab.distlcpvalues,NULL);
gt_disc_distri_delete(outlcpinfo->lcpsubtab.distlcpvalues);
}
gt_free(outlcpinfo);
}
void gth_stat_delete(GthStat *stat)
{
if (!stat) return;
gt_disc_distri_delete(stat->exondistribution);
gt_disc_distri_delete(stat->introndistribution);
gt_disc_distri_delete(stat->matchnumdistribution);
gt_disc_distri_delete(stat->refseqcoveragedistribution);
gt_disc_distri_delete(stat->sa_alignment_score_distribution);
gt_disc_distri_delete(stat->sa_coverage_distribution);
gt_free(stat);
}
static void stat_visitor_free(GtNodeVisitor *nv)
{
GtStatVisitor *sv = stat_visitor_cast(nv);
gt_cstr_table_delete(sv->used_sources);
gt_disc_distri_delete(sv->cds_length_distribution);
gt_disc_distri_delete(sv->intron_length_distribution);
gt_disc_distri_delete(sv->exon_number_distribution);
gt_disc_distri_delete(sv->exon_length_distribution);
gt_disc_distri_delete(sv->gene_score_distribution);
gt_disc_distri_delete(sv->gene_length_distribution);
}
static inline GtDiscDistri *read_zero_disc_distri(GtDiscDistri *dist,
FILE *fp, GtError *err)
{
int had_err = 0;
GtUword idx, symbol;
GtUint64 freq;
GtUword num_of_zero_leaves;
dist = gt_disc_distri_new();
had_err = gt_io_error_fread_one(num_of_zero_leaves, fp, err);
for (idx = 0; !had_err && idx < num_of_zero_leaves; idx++) {
had_err = gt_io_error_fread_one(symbol, fp, err);
if (!had_err)
had_err = gt_io_error_fread_one(freq, fp, err);
if (!had_err)
gt_disc_distri_add_multi(dist, symbol, freq);
}
if (had_err) {
gt_disc_distri_delete(dist);
dist = NULL;
}
return dist;
}
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 inline GtDiscDistri *write_zero_disc_distri(GtDiscDistri *dist,
EncsecDistriData *data,
GtError *err)
{
GtUword num_of_zero_leaves;
data->written_elems = 0;
data->err = err;
data->had_err = 0;
gt_disc_distri_foreach(dist, encdesc_distri_iter_count, data);
if (!data->had_err) {
num_of_zero_leaves = data->written_elems;
}
data->had_err = gt_io_error_fwrite_one(num_of_zero_leaves, data->fp, err);
if (!data->had_err) {
data->written_elems = 0;
gt_disc_distri_foreach(dist, encdesc_distri_iter_write, data);
gt_assert(data->written_elems == num_of_zero_leaves);
}
if (data->had_err) {
gt_disc_distri_delete(dist);
dist = NULL;
}
return dist;
}
