static void gt_pbs_add_hit(GtArray *hitlist, GtAlignment *ali, GtPBSOptions *o,
unsigned long trna_seqlen, const char *desc,
GtStrand strand, GtPBSResults *r)
{
unsigned long dist;
GtPBSHit *hit;
unsigned long offset;
GtRange urange, vrange;
gt_assert(hitlist && desc);
if (!ali) return;
gt_assert(ali);
dist = gt_alignment_eval(ali);
urange = gt_alignment_get_urange(ali);
vrange = gt_alignment_get_vrange(ali);
offset = abs(o->radius - urange.start);
if (dist <= o->max_edist
&& abs(o->radius-urange.start) <= o->offsetlen.end
&& abs(o->radius-urange.start) >= o->offsetlen.start
&& abs(urange.end-urange.start+1) <= o->alilen.end
&& abs(urange.end-urange.start+1) >= o->alilen.start
&& vrange.start <= o->trnaoffsetlen.end
&& vrange.start >= o->trnaoffsetlen.start)
{
hit = gt_pbs_hit_new(abs(urange.end-urange.start+1),
strand,
desc,
vrange.start,
urange.start,
urange.end,
offset,
dist,
gt_pbs_score_func(dist,
offset,
urange.end-urange.start+1,
trna_seqlen,
vrange.start),
r);
gt_array_add(hitlist, hit);
}
}
static GtMatchIteratorStatus gt_match_iterator_sw_next(GtMatchIterator *mi,
GT_UNUSED GtMatch **match,
GT_UNUSED GtError *err)
{
GtMatchIteratorSW *mis;
GtSeq *seq_a, *seq_b;
char *a, *b;
const char *adesc, *bdesc;
GtAlignment *ali = NULL;
unsigned long seqlen_a, seqlen_b, seqpos;
GtRange arng, brng;
gt_assert(mi && match);
mis = gt_match_iterator_sw_cast(mi);
while (true) {
if (!mis->pvt->firstali)
mis->pvt->seqno_es2++;
if (mis->pvt->seqno_es2 == gt_encseq_num_of_sequences(mis->pvt->es2)) {
mis->pvt->seqno_es1++;
if (mis->pvt->seqno_es1 == gt_encseq_num_of_sequences(mis->pvt->es1))
return GT_MATCHER_STATUS_END;
mis->pvt->seqno_es2 = 0;
}
seqlen_a = gt_encseq_seqlength(mis->pvt->es1, mis->pvt->seqno_es1);
seqlen_b = gt_encseq_seqlength(mis->pvt->es2, mis->pvt->seqno_es2);
/* XXX: reuse buffers for performance improvement */
a = gt_malloc(seqlen_a * sizeof (char));
seqpos = gt_encseq_seqstartpos(mis->pvt->es1, mis->pvt->seqno_es1);
gt_encseq_extract_decoded(mis->pvt->es1, a, seqpos, seqpos + seqlen_a - 1);
b = gt_malloc(seqlen_b * sizeof (char));
seqpos = gt_encseq_seqstartpos(mis->pvt->es2, mis->pvt->seqno_es2);
gt_encseq_extract_decoded(mis->pvt->es1, b, seqpos, seqpos + seqlen_b - 1);
seq_a = gt_seq_new(a, seqlen_a, gt_encseq_alphabet(mis->pvt->es1));
seq_b = gt_seq_new(b, seqlen_b, gt_encseq_alphabet(mis->pvt->es2));
ali = gt_swalign(seq_a, seq_b, mis->pvt->sf);
mis->pvt->firstali = false;
if (ali && gt_alignment_get_length(ali) >= mis->pvt->min_len
&& gt_alignment_eval(ali) <= mis->pvt->max_edist) {
break;
}
gt_alignment_delete(ali);
gt_seq_delete(seq_a);
gt_seq_delete(seq_b);
gt_free(a);
gt_free(b);
}
arng = gt_alignment_get_urange(ali);
brng = gt_alignment_get_vrange(ali);
adesc = gt_encseq_description(mis->pvt->es1, &seqlen_a, mis->pvt->seqno_es1);
bdesc = gt_encseq_description(mis->pvt->es2, &seqlen_b, mis->pvt->seqno_es2);
*match = gt_match_sw_new("", "",
mis->pvt->seqno_es1,
mis->pvt->seqno_es2,
gt_alignment_get_length(ali),
gt_alignment_eval(ali),
arng.start, brng.start,
arng.end, brng.end,
GT_MATCH_DIRECT);
gt_match_set_seqid1_nt(*match, adesc, seqlen_a);
gt_match_set_seqid2_nt(*match, bdesc, seqlen_b);
gt_alignment_delete(ali);
gt_seq_delete(seq_a);
gt_seq_delete(seq_b);
gt_free(a);
gt_free(b);
return GT_MATCHER_STATUS_OK;
}
